Digital PDFs

AA-N149A-TC

May 1982

124 pages

Original

31MB

view download

OCR Version

12MB

view download

Document:	DECnet Digital Network Architecure (Phase IV) General Description
Order Number:	AA-N149A-TC
Revision:	0
Pages:	124
Original Filename:

OCR Text

////

,///

S
s
/////////
//S//////////////////

dliigliltlall

Trton [ mnn

DECnet

- DIGITAL Network Archltecture._j_ |
~(Phase IV)
|

General Descnptlon
‘Order No. AA-N149A-TC

May 1982

This document describes the design of the DIGITAL Network Architecture that serves as a model for Phase |V DECnet implementations. It

includes descriptions of functions, protocol messages, and operation.

To order additional copies of this document, contact your local
Digital Equipment Corporation Sales Office.

v digital equipment corporation - mognard, massachusetts

First Printing, May 1982

'This material may be copied, in whole or in part, provided that the copyrlght notice belowisincluded
in each copy along with an acknowledgment that the copy descr1bes the DigitalNetwork Architecture
|

developed by Digital Equipment Corporatlon

This material may be changed without notice by Dlgltal Equ1pment Corporatlon and Dlgltal
Equlpment Corporationis not responsibie for any errors which may appear herein.

The postage—prepaid READER’S COMMENTS form on the last page of this document requests the
user’s critical evaluation to assist us in preparing future documentation.

VAX
- DECnet

DATATRIEVE

DECsystem-10
.»
DECtape
DIBOL
EDUSYSTEM
\
FLIP CHIP
|
FLOCAL
INDAC
.
LAB-8
DECSYSTEM-20
RTS-8
VMS
IAS

TRAX

MASSBUS
OMNIBUS
- 05/8
PHA
RSTS
RSX
TYPESET-8
TYPESET-11
TMS-11
- ITPS-10

|
|

SBI
PDT

DIGITAL
~
|
DEC
PDP
DECUS
UNIBUS
COMPUTER LABS
COMTEX
’
|
DDT
DECCOMM
ASSIST-11

The following are trademarks of Digital Equipment Corporation:

Contents
Page

Preface
Chapter1

1.2
1.3

Layers, Modules, and Interfaces
User Services
=
)

1.5

DataFlow

1.4

Protocols

1-3
B

2.1.1
DDCMP Messages
|
|
2.1.2
DDCMP Operation
2.2 Ethernet Functional Description
2.2.1

Ethernet Messages

'2.2.2

Ethernet Operation

1-8

1-11

|
-

|
|
o

2.3 X.25Functional Description
. 2.3.1
X.25Frame Level
© 2.3.2 X.25 Packet Level

2-3
2-6
2-10
-2-11

2-1

2-13
S

2-15
2-15
2-16

The Routing Layer

- 3.1 Routing Functional Descriptio‘h
3.2 Routing Messages

3.2.1

Packet Route Header =

3-1

|
|

3.2.2 Routing Control Messages o
R
3.3 Routing Operatlon
|
|
o
3.3.1
Routing
S
|
S
B
- 3.3.2 Congestion Control
o
-

3.3.3
3.3.4

Chapter 4

1-4
1-8

The Data Link Layer

2.1 DDCMP Functional Description

Chapter 3

vii

Introduction

1.1 Design Goals

Chapter2

Packet Lifetime Control ~
[Initialization and Circuit Monitor

|
»

3-5
3-8
3-8
3-8

3-9
39

o
|

3-4

TheEnd Communication Layer
4 1 NSP Functmnal Descrlptlon
4.2
4.3

NSP Messages
NSP Operation
4.3.1
4.3.2
4.3.3
4.3.4

S
|
|
-

Logical Links
o
Segmentation and Reassembly of Data
Error Control
|
A
‘
Flow Control

4-1

43
4-3

- 4-5
4-5
4-7

Chapter 5

The Session Control Layer
5.1

Chapter 6

Session Control Functional Description

5-1

5.2

Session Control Messages

5-2

5.3

Session Control Operation
Requesting a Connection

5.3.2

Receiving a Connect Request

5-3

5.3.3

Sending and Receiving Data

5.3.4

Disconnecting and Aborting a Loglcal Lmk

5.3.5

Monitoring a Logical Link

5-4

The Network Applicatidn Layer
6.1

6.2

6.3

6.4

Chapter?7

5.3.1

DAP Functional Description
6.1.1

DAP Messages

6.1.2

DAP Operation

6.1.3

DECnet Remote File Access Fac1ht1es

6-2

6-5

Network Virtual Terminal Functional Description

6.2.1

The Terminal Communication Protocol

6.2.2

The Command Terminal Protocola -

6.2.3

NVT Operation

6-7

6-9

6-11

X.25 Gateway Access Functlonal Descrlptmn

6-13

6.3.1

X.25 Gateway Access Messages

6-13

6.3.2

X.25 Gateway Access Operation

6-14

6.3.3

X.25 Gateway Access Modules

6-16

SN
A Gateway Access Functional Deseription

6-18

6.4.1

SNA Gateway Access Protocol Messages

6-18

6.4.2
6.4.3

SNA Gateway Access Operatior} |
SNA Gateway Access Modules

619

6-20

Network Management
7.1

Network Management Functional Description

7.2

Network Management Operation
7.2.1

Components

7.2.2

Remote Loading, Dumpmg, Controlhng, and Link Loopback

7-2

7-2
s

Testing Functions

7.2.3

7.2.4

Glossary

Node Loopback Testmg

Parameters, Counters, and Events

7.3

Network Management Messages

7.4

Maintenance Operation Functional Descrlptlon

7-8
7-11

7-13

7.41

MOP Messages

7-13

7.4.2

MOP Operation

- 7-14

F_ig’u res
1-1

Basic DNA Structure

1-2

DNA

1-3

DNA Modules Resident in a Typical DECnet Node

1-2

Layers and Interfaces

1-6

1-4

Protocol Communication between Two Nodes

1-5

Information Building as Data Traverses DNA Layers

1-11

1-6

Data Flow

1-14

2-1

DDCMP Data Message Format

2-3

DDCMP Control Message Formats

2-5

2-3

DDCMP Maintenance Message Format

2-6

2-4

Typical DDCMP Message Exchange, Showing Positive

- 2-2

1-10

Acknowledgment, Piggybacking, and Pipelining
2-5

DDCMP Message Exchange, Showing Error Recovery

2-6

Ethernet Data Link Layer Functions

2-1

M,’/

2-8

2-9

2-11

Ethernet Data Link Frame Format

2-12

Ethernet Frame Format with Padding

2-13

3-1

Routing Terms

3-2

Packet Route Header Formats

3-3

3-5

3-3

Routing Control Message Formats

3-4

Routing Layer Components and their Functions

4-2

Typical Message Exchange Between Two Implementations of NSP

4-4

4-3

Acknowledgment Operation

4-6

Logical Links

3-7

3-11

4-4

Segment Flow Control Shown
in One Direction on a Logical Link

4-8

5-1

A Session Control Model

5-2

Session Control Message Formats

6-1

DAP Message Exchange (Sequential File Retrieval)

6-4

File Transfer Across a Network

6-6

6-3

Network Virtual Terminal Service

6-8

6-4

NVT Message Exchange

6-12

6-5

X.25 Gateway Access Message Exchange

6-15

X.25 Gateway Access Operation

6-17

SNA Gateway Access Message Exchange

6-20

6-8

SNA Gateway Access Operation

6-21

7-1

Relationship among Network Management Components at a

6-6

Single Node

7-2

Down-Line Load Request Operation

7-3

Line Loopback Tests

7-4

Examples of Node Level Testing Using a Loopback Node
Examples of Node Level Logical Link Loopback Tests

7-4
7-6
7-9
7-10

Tables
X.25 Level 2 Frame Types

2-16

2-2

X.25 Level 3 Packet Types

2-18

NSP Messages

4-2

6-1

DAP Messages

6-3

Terminal Communication Protocol Messages

6-9

6-10

6-3

Command Terminal Protocol Messages

6-4

X.25 Gateway Access Messages

614

SNA Gateway Access Messages

6-19

7-1

NICE Messages

7-12

7-2

Loopback Mirror Messages

7-12

7-3

MOP Messages

7-13

Preface
This document is an overview of the DIGITAL Network Architecture (DNA). DNA is a model
of structure and function upon which DECnet implementations are based. DECnet is a family
of communications software and hardware products that enable DIGITAL operatmg systems
and computers to functionin a DECnet network.
A DECnet network is a group of DIGITAL computer systems with associated operating
systems, DECnet software, and communication hardware that are connected to each other by
physical channels or lines. Each computer in a network containing a DECnet implementation
is called a node. The network therefore consists of connected nodes and lines. DNA defines

standard protocols, interfaces, and functions that enable DECnet network nodes to share data
and access each others’ resources, programs, and functions.

This document describes Phase IV DNA. Phase IV DECnet implementations will include
DECnet-VAX, DECnet-11M, DECnet-11M-PLUS and DECnet-11S, Over time, Phase IV
support will be provided for DECsystem-20 and the Professional 3XX Series. The previous
version of DECnet, Phase III, is compatible with Phase IV. However, Phase III nodes can
provide Phase III functions only.

DNA supports a broad range of applications and a variety of network topologies. (A network
topology is a particular configuration of nodes and lines.) User documentation and marketing
brochures describe in detail various types of applications and specific programming and
network management information.

This document summarizes the design and structure of DNA and serves as an introduction to
the DNA functional specifications. It is intended for readers with a knowledge of
communications technology who desire an understanding of the overall DNA structure. A
glossary at the end of the document defines many DNA terms. Additionally, many DNA
terms are italicized and explained in the text at their first occurrence.

The DNA functional spec1ficat1ons containing the archltectural details of DNA, include the
|
following:
DNA Data Access Protocol (DAP) Functional Speczficatwn Version 5.6.0, Order No.
- AA-K177A-TK

DNA Digital Data Communications Message Protocol (DDCMP) Functional Specification,
Version 4.1.0, Order No. AA-K175A-TK

DNA Session Control Functional Specification, Version 1.0.0, Order No. AA-K182A-TK
Functional specifications for the following layers of DNA Phase IV are not currently
available, but will be available c01nc1dent with the the avallablhty of DECnet-VAX Phase IV
support

DNA Routing Functional Specification,. Phase IV
DNA Maintenance Operations Functional Specification, Phase IV

DNA Network Managemént Functional Specification, Phase IV
DNA End Commaunications Functional Specification, Phase IV

Vil

The following specifications describe functions supported by DNA Phase IV: |
The Ethernet, A Local Area Network, Data Link Layer and Physzcal Layer Speczficatton
Version 1.0, Order number AA-7 59A TK

CCITT Recommendation X.25, Interface between Data Terminal Equzpment and Data
Circuit Terminating Equlpment for Termznals Operating in the Packet Mode on Publtc

Data Networks

viil

User

Network Management
Network Application

Session Control
End Communication
Routing

Data Link

Physical Link

Chapter 1

Introduction
A network architecture specifies common communication mechanisms and user
interfaces that computer systems of different types must adhere to when passing data
between systems.
A network architecture may also be designéd so that
implementations meet other goals, such as cost effectivness and flexibility.
A network architecture is necessary for several reasons:

Communications technology is changing continually. It is necessary to have a
structure for networking that can adjust to these changes as effectively
as
possible.

A variety of operating systems, communications devices, and computer hardware

exists.

A common standard of networking to which each operating system or

:\ \\"-_-///’

communication hardware facility can adhere is necessary.

A network, to be most useful and cost effective, should provide a broad range of
userapplications and functions. This requires very complex software. A network
architecture forces such software to have a clean, well-structured design.

Network management, error recording, and maintenance are easier when
implementations provide standard procedures for error detection, recording,
isolation, recovery, and repair. An architecture provides this standardization. -

Networks need to be adaptable to different communication situations. A
hierarchical, modular architecture enables the substitution of modules of
equivalent function to suit the specific needs of a particular network. For
example, a data link protocol that ensures data integrity over a leased physical
circuit can be replaced by a protocol that allows two systems to communicate over
a public data network.

A network architecture enables system-independent functions to be designed
specifically enough to be implemented in hardware. This can improve system
performance. For example, the DEUNA communications device implements the

Ethernet data link, the DMR11, DMP11 and DMV11 implement the DDCMP

data link, and the KMS11-BD implements the X.25 frame level. Basically,
DIGITAL Network Architecture is the specification of a layered hierarchy in

Introduction

1-1

which each

layer contains modules

that perform defined functions.

The

architecture specifies the functions of these modules and relationships between

them.

Modules in a layer typically use the services of modules in the layer immediately
below. Some layers may contain more than one type of module, but in this case the
modules fall into a more general category of function.

For example, user-written

~ programs all operate in the top layer, or User layer. The bottom layer is the most
basic layer. It manages the physical transmission of information over a channel.
Each module specified by DNA operates as a black box. That is, the operation within
the black box is transparent to other layers and to equivalent modules in other nodes.

The architecture does not define specific code for implementing modules. It only
defines specific operations that the modules must perform.

These definitions may

take the form of algorithms written in a higher level programming language.

The architecture specifies two kinds of relationships between modules:
®

Interfaces. Interfaces are the relationships between different modules that are

usually in the same node.

Typically, a module in one layer interfaces with a

module in the layer immediately below to receive a service and with a module in

the layer immediately above to provide a service. The architecture specifies the
functions supplied by these interfaces as calls to subroutines. These specifications

are functional: they need not be implemented as subroutine calls.

Protocols. Protocols are the relationships between equivalent modules that are
usually in different nodes.

These protocols consists of messages with specific

formats and rules for exchanging the messages.

Such protocols are called peer

protocols.

A state table is often specified to show the transitions occurring in a black box when
protocol message exchanges or interface calls from higher layers take place.
Figure 1-1 shows the basic architectural structure of DNA.

MODULES

PROTOCOLS

\4
)
|
MODULES | <«—ROTOOL
¢ INTERFACES

|
|

MODULES | «—T"210COL
,L INTERFACES

MODULES | «——2210%0
Node A

MODULES

A
-

Layer3
¢ INTERFACES

o | MODULES |
rayer2
; INTERFACES
|

mopuLes
Node B |

Figure 1-1: Basic DNA Structure

1-2

Introduction

Layern

Layer

The architecture specifies common error reporting, operational parameters, and
counters

that certain layers must maintain.

This standardization ensures that

maintenance, error logging, and network management can take place consistently

and, to some extent, across systems.

For example, an operator could manage an

unattended remote system from his local node. He could expect to observe or, in some
cases, change values of parameters equivalent in function and name to those at his
own node, even if the remote node was running under a different operating system.
DNA is an open-ended architecture.

Future phases of DNA may model additional

layers, additional modules within existing layers, or alternative models for certain
layers.

1.1

Design Goals

DNA achieves the following design goals:
Create a common user interface.

The application interface to the network is

common across the varied implementations. The common interface hides the internal
characteristics and topology of the network.

Support a broad spectrum of applications. DECnet networks can:

® Share resources.
®

Distribute computation.

® Communicate efficiently with remote systems.
Resource-sharing activities include remote file access, inter-computer file transfer,
distributed data base queries, task-to-task communication, and remote use of

peripheral devices.

Data moved across the network can be streams of related

sequential data and short independent messages.

Distributed computation means cooperating programs

computers in a network.

executing in

different

Examples are real-time process control (such as a

manufacturing system), specialized data base multiprocessor systems (such as orderprocessing or payroll systems), and distributed processing systems (such as a banking,

airline reservation, or combined order-processing/inventory control system).

Remote communication facilities include remote ‘public data network interfaces,
Ethernet interactive terminals, and batch entry/exit stations.
Support a wide range of communication facilities.

Such facilities include a

variety of asynchronous and synchronous, full- and half-duplex communications
devices, some with multipoint (multidrop) and/or multiple controller/multiplex
capabilities.

(Consult the glossary for definitions.)

DECnet networks support a

~ variety of communication channels, such as leased lines, satellite links, Ethernet
local area networks, X.25-based packet switching networks, and local links.

Be cost effective.

A DECnet network application costs about the same as a custom-

designed network that achieves the same performance for the same application.

Support a wide range of topologies.

DECnet networks support

Introduction

many

1-3

These range from point-to-point, star, or bus
topologies to hierarchical structures to topologies in which each node has equal
control over network operation.
Be highly available.

~—_”

- configurations of nodes and lines.

DECnet networks can be configured to maintain operation

even if a subset of lines or nodes fail.

Because maintenance functions are highly
distributed, DECnet networks can recover from operator error unless the error is
extreme.
-

Be extensible.

DNA allows for incorporation of future technology changes in

hardware and/or software. Moreover, DNA makes possible the movement
of functions

from software to hardware. DNA also permits subsets of modules.
Be highly dlstrlbuted The major functions of DNA are not centralizedin one node
in a network.

Implement network control and maintenance functions at a user level.

permits easier system management.

This

For example, termmal commands can set,

change or display lower level parameters or counters.
‘Allow for security.

The DNA design allows for security at several levels.

User

access control and the authentication of nodes is implemented in most DECnet

products.

1.2 Layers, Modules, and Interfaces
DNA defines the following layers, described in order from the highest to the lowest:

User layer. The User layer contains most user-supplied functions. It also contains
the Network Control Program, a Network Management module that gives system
managers interactive terminal access to lower layers.

Chapter 7 describes the

function of the Network Control Program, the only DNA-defined User layer module.
Network Management layer. Modules in the Network Management layer provide
user control of and access to operational parameters and counters in lower layers.

Network Management also performs down-line loading, up-line dumping, remote
system control, and test functions. In addition, Network Management performs event
logging functions. This layer is the only one that has direct access to each lower layer

Chapter 7 describes Network Management.
Network Application layer.

services to the User layer.

The Network Application layer provides generic

Services include remote file access, remote file transfer,

remote interactive terminal access, gateway access to non-DNA systems, and resource
managing programs. This layer contains both user- and DIGITAL-supplied modules.

Modules execute simultaneously and independently in this layer.
The Network
Application layer contains functions of both the Application and Presentation layers

of the ISO Reference Model. Chapter 6 describes the Network Application layer.
Session Control layer.

1-4

Iritroduction

a physical one) on which information flows in two directions.
Session Control
functions include name to address translation, process addressing, and, in some

The Session Control layer defines the system-dependent
aspects of logical link communication. A logical link is a virtual circuit (as opposed to

systems, process activation and access control. The Session Control layer corresponds
to the Session layer of the ISO Reference Model.

Chapter 5 describes the Session

Control layer.

End Communication layer (previously called the Network Services layer). The End

Communication layeris responsible for the system-independent aspects of creating
and managing logical links for network users. End Communication modules perform
data flow control, end-to-end error control, and the segmentation and reassembly of

user messages. The End Communication layer corresponds to the Transport layer of
the ISO Referennce Model. Chapter 4 describes the End Communication layer.
Routing layer (previously called the Transport layer). Modules in the Routing layer
route user data, called a datagram and contained in a packet, to its destination.

Routing modules also provide congestion control and packet lifetime control. The
Routing layer corresponds to the Network layer of the [ISO Reference Model. Chapter

3 describes the Routing layer.

Data Link layer.

Modules in the Data Link layer create a communication path

between adjacent nodes. Data Link modules ensure the integrity of data transferred
across the path. Data link modules for Ethernet local area networks, X.25 public data
networks, and synchronous or asynchronous lines execute simultaneously and

independently in this layer. Chapter 2 describes the Data Link layer.
Physical Link layer.

Modules in the Physcial Link layer manage the physical

transmission of data over a channel. The functions of modules in this layer include

monitoring channel signals, clocking on the channel, handling interrupts from the
hardware, and informing the Data Link layer when a transmission is completed.

Implementations

this

layer

encompass

parts

of device

drivers

for

each

communication device as well as the communication hardware itself. The hardware

includes devices, modems, and lines. In this layer, industry standard electrical signal
spec1ficat10ns such as EIA RS-232C, or CCITT V .24, the Ethernet physical layer, or
CCITT X.25 level 1 operate rather than peer protocols. Thereis no chapter devoted to
this layer since only its Network Management interface, counters, and events are
DNA defined.

Figure 1-2 shows the relationships among the DNA layers in a single node. The three

upper layers each interface directly with Session Control for logical link services.
Each layer interfaces with the layer directly below to use its services. The User layer

interfaces directly with the Network Application layer as well. In addition, Network

Management modules interface directly with each lower layer for access and control
purposes. Finally, Network Management interfaces directly w1th the Data Link layer
for service functions that do not require logical links.

Figure 1-3 shows a typical DECnet node containing multiple modules in some layers.

Introduction

1-5

/
\

User Modules

]

Network |Management Modules

Network Application

Modules

Session Control Modules

End Communication Modules J

Routing

Modules

Data Link

Physical

Modules

Link

Modules

[ ,

Interface Key

Clent

Network

1-6

Introduction

> Management

Figure 1-2:

DNA Layers

and Interfaces

The Network Control
Program, which allows
manager/operator to

@ A user program
communicating with a
remote program

monitor/control

® A user program
accessing remote files

network activity

\\\\* o

N -~ ~< -

\\\

User

S ~

NCP

Layer

Utility

‘\1

}

User ®

l’,

User @

Program

Network

. P

Network

Management

Management
Layer

Routines
AN

Network

Remote

File Access

Application
Layer

Routines

Session

Session Control

Control

Module

Layer
v

End

NSP

Communication
Layer

Module

|
A 4

Routing

Layer

/MOdu'e\

Data

DDCMP

Link

Module

Layer

P.h ysica

Line A’s

. 'E' nk

Controller

Line B’s

xa2s5

Line C's

Controller | | Controller

Communications

Facilities

Ethernet

Module

Line D’s

Controller

Line E's

Controller

V]
Line A(e.g. 2

telephone line)

LineB(e.g.a
local cable)

LineC{e.g.a
_
telephone line)

|
(Ethernet cables)

Figure 1-3: DNA Modules Resident in a Typical DECnet Node

Introduction

1-7

1.3 U_ser Services _
As shown in Figures 1-2 and 1-3, there are different ways in which modules use the
-~

services of other modules in a system. DNA allows the following services:

® User-to-user. A user level process in one node communicates via a logical link

with a user level process in another node. For example, a FORTRAN call from the
user level might, using a direct interface to the Session Control module that
communicates with a remote node, send data to a user level process in another
node.

Remote application. A user level process uses a Network Application module
which, in turn, uses a logical link to perform a function at a remote node. For
example, a user level command causes a Network Application module to transfer
a file to a remote node. The Network Application module uses a protocol called
- DAP to communicate with the remote Network Application module.

Network Management. A user level command or process uses a Network
Management module to perform a Network Management service at a remote
node. Network Management performs some services using a logical link provided
by Session Control and other services by interfacing directly to the Data Link
layer. An example of a Network Management operation is a command that
displays a remote node’s counters at a local terminal.

1.4 Protocols
Modules with equivalent functions in the same layer, but in different nodes,
communicate using protocols. A protocol is both a set of messages and the rules for
exchanging the messages. The architecture defines the message formats and rules
very specifically.
|

Protocols are necessary for every DNA function that requires communication between
~

nodes. For example, DDCMP, which resides in the Data Link layer, assigns numbers

to transmitted messages and checks the numbers of received messages. In this way,

the protocol ensures that transmitted data is received in the correct order.

The DN A-defined protocols, listed accordirig to layer, are:
Network Management Layer
®

Network Information and Control Exchange (NICE) Protocol. This is used
for triggering down-line loading, up-line dumping, testing, reading parameters
and counters, setting parameters, and zeroing counters.

Event Logger Protocol. This is used for recording significant occurrences in

lower layers. An event could result from a line coming up, a counter reaching a
threshold, a node becoming unreachable, and so on.

Maintenance Operation Protocol. This protocol performs data link level
loopback tests, remote control of unattended systems, and down-line loading/up-

1-8

Introduction

line dumping of computer systems without mass storage.

Network Application Layer

'~ ® The Data Access Protocol (DAP). This is used for remote file access and
transfer.

® - The Network Virtual Terminal Protocols. This family of protocols is used for
terminal access through the network.

X.25 Gateway Access Protocol. This protocol allows a node which is not
connected directly to a public data network to access the facilities of that network
through an intermediary gateway node.

SNA Gateway Access Protocol. This protocol allows a node which is not
connected directly to an IBM SNA network access to the facilities of the SNA
network for terminal access and remote job entry.

® The Loopback Mirror Protocol. This is used for Network Mana‘gement logical
link loopback tests.

Session Control Layer
®

Session Control Protocol.

This is used for functions such as sending and

receiving logical link data, and disconnecting and aborting logical links.

End Communication Layer

° Network Services Protocol (NSP). This handles all the system-independent
aspects of managing logical links.

“Routing Layer
®

Routing Protocol. This handles routing and congestion control.

Data Link Layer
®

Digital Data Communications Message Protocol (DDCMP).

This ensures

integrity and correct sequencing of messages between adjacent nodes.

- ®

X.25 Protocol. This implements the X.25 packet level (level 3) and X.25 frame

level (level 2) of the CCITT X.25 recommendation for pubhc data network
1nterfaces

Ethernet Protocol This implements the Ethernet data link protocol for
communication between adjacent nodes connected by an Ethernet local area
network.

Figure 1-4 shows protocol communication between two nodes. Some implementations

may use the Network Services Protocol (NSP) that creates and manages logical links
for network communication within a single node, as well as for inter-node

communication. For example, Network Management modules may use a logical link
~ even when performing local functions.

Introduction

1-9

NODE §
MODULES

NODE 2
MODULES

PROTOCOLS
User Layer

User Program

USER-DEFINEDPROTOCOL

User Program

Network Management Layer

Module

Network Application Layer

Management -

Module

File Access

Network File
Access Routines
(NFARsS) .

Network

NETWORK INFORMATION AND
CONTROL EXCHANGE
PROTOCOL (NICE)

DATA ACCESS PROTOCOL (DAP)

MEEND

WRER

GOUS

WEEN

Management

Network

Listener
(FAL)

SESSION CONTROL PROTOCOL

Session Control
Module

Module

NSP Module

Session Control

Session Control Layer

Routing Module

NSP Module

End Communication Layer

NETWORK SERVICES PROTOCOL
(NSP)

Routing Layer

ROUTING PROTOCOL

Routing Module

Data Link Layer
DIGITAL DATA

DDCMP Module

COMMUNICATIONS MESSAGE PROTOCOL (DDCMP)

Device

ELECTRICAL SIGNALS

Controller

Physical Link Layer
Dewvice
Controller

J Communication lines l

forming physical
connection

Figure 1-4: Protocol Communication between Two Nodes

1-10

Introduction

1.5 _Da_ta Flow
' The primary purpose of a network is to pass data from a source in one node to a
destination in another. Because DNA is layered, it is important to understand how
data flows through these layers and between nodes. Data traveling from one node in a

network to another passes from a source process in the user layer down through each

layer of the DNA hierarchy of the source node before being transmitted across a line.
If the destination node is not adjacent, the data must then travel up to the Routing
layer of the adjacent node, where it is routed (or switched), sent back down through
the two lower layers, and transmitted across the next line in the path. The data keeps
travelling in this manner until it reaches its destination node. At this node, the data

passes up the hierarchy of layers to the destination process.

Figure 1-5 shows how information is built as data passes through the DNA layers at

one node. In this example, Network Management is not involved.

User

Considered as data
by layer

Layer

Network
Application
Control
Information

~ Network

Application
Layer

Session
Control
Control
Information

Session
Control
Layer

End

|}
Control.
Information |

Layer

Routing

Roufing

Control
Information

Layer

Link
Layer

End
Comm.

Communication

Data

User

Data

Link
Control
Information

Link
Control
information
v

Complete Envelope Transmitted
from One Node to an Adjacent Node

Figure 1-5: Information Building as Data Traverses DNA Layers

Introduction

1-11

In the following scenario a user attempts to form a logical link with another user. The

requesting user passes initial connection data to the destination user. The numbered
steps correspond with numbers on Figure 1-6 which follows this explanation.

Data Flow at the Source Node
@ ~ The source user requests a connectmn to the destmatlon user and passes connect
data.

The Session Control module receiVes the data, maps the destination node name
to a numerical address, if necessary, and places the data in the next transmit
buffer, adding control information to the message The message then passes to
‘the End Communication layer.

The End Communication module adds its control. information (including a
logical link identification) and passes the message (now called a datagram) to
the Routing layer.

The Routing module adds a header consisting of the destination and source node

addresses and selects an outgoing circuit for themessage based on routing
information. Routing then passes the message (now called a packet) to the Data

Link layer, specifying the outgoing circuit address and, if necessary, the station
address of the receiving node on the circuit.

The Data Link module adds its protocol header consisting of framing,
synchronizing, addressing and control information, and its protocol trailer

consisting of a cyclic redundancy check (CRC). The packetis now enveloped for
transmlssmn

The Physical Link module transmits the enveloped message over the physwal
line.

Data Flow Across the Network to the Destination Node

The enveloped data message arrives at the next node. The Physical Link module
- receives the message and passes it to the Data Link layer.

The Data Link module checks the packet for bit errors in transmission. On links
with a significant probability of transmission errors, the Data Link protocol

performs error correction.
retransmission.

DDCMP and X.25 provide error correction by

On links with a low probability of transmission error, for

example Ethernet links, packets received in error are discarded, with error
recovery provided by higher layers.

The Data Link header and trailer are

removed from a correctly received packet whichis then passed up to the Routing
layer.

- The Routing layer checks the destination addressin the header. If the addressis
not this node, Routing selects the next outgoing circuit from its routing table,
and passes the message back to the Data Link layer.

The Routing layer has

routed the message on to the next line in its path. The message proceeds as in
steps 5 and 6 above.

1-12

Introduction

The message proceeds through the network, switching at routing nodes, until it

arrives at the Routing layer with the same address as the destination addressin
the message

Data Flow at the Destination Node
@

The packet passes to the Routmg layer of the destination node as descrlbedin 7
and 8 above. The destination Routing module removes the Routmg header, and
passes the datagram to End Communication.

The End Communication layer module examines the End Communication
layer
header on the datagram.

If it has the resources to form a new logical link, it

‘passes the connect data,

without the End Communication layer control

information, to Session Control.

The Session Control module performs any necessary access control functions and
passes the message to the appropriate process in the user layer after removing
Session Control header information.
The destlnatlon process interprets the data according to whatever higher level
protocolis bemg used.

Figure 1-6 illustrates the data flow just described.

Introduction

1-13

1-14

Introduction

sokeq13pedH19pedH~.JET]-ETW

3@ s13peaHpueelep!s0i13wpbeoscH€pueeje:p||Ai

1uje9o5l8fneNeqI|UNW[0eONloUeD0q))|BJ@eu1yn9epeoayyuij'13he|dpouvvii‘|vvad|0e4lUeq0us)l1e3.,hyeyp0y1ualy0)‘}1b33fupe[OnUodp.-Zo“u19"A2e|93S)24n\Fb0Re13ru4eR9(pnp-ooy1uuyi|‘£'13Ae|9BPOUj‘oZesl¥ueq0)|Bius@pnueiaeoHyye|p@yui|‘13he|vvd|i|ipvvoaau"€j0eIj3Ue0qD)
:9-19angdiyereqMol

e
J
Y
p
[
1
U
P
u
e
0
a
O
I
4
A
J
r
}
e
i
3
p
o
1
I
l
U
O
u
w
0
q
a
j
W
|
)
O
e
d
.
I
<S01_5mSPU13OAI5PoObBW©IWO:AJUHJI

User

Network Management

Network Application
Session Control
End Communication
Routing

Physical Link
.....

Chapter 2

The Data Link Layer

The Data Link layer, residing immediately above the Physical Link layer, is
responsible for creating a communications path between adjacent nodes. The Data
Link layer frames messages for transmisssion on the channel connecting the nodes,
- checks the integrity of received messages, manages the use of channel resources, and,

when required, ensures the integrity and proper sequence
of transmitted data.

Currently there are three protocols residing in the DNA Data Link layer:

Digital Data Communications Message Protocol (DDCMP).

DDCMP

operates over synchronous or asynchronous communications links. It provides
correct sequencing of data and error control to ensure data integrity. Section 2.1
describes DDCMP.

Ethernet Data Link. The Ethernet Data Link operates over a coaxial cable
based local area network. It provides a best-effort delivery service and includes
link management procedures and error detection capability. Section 2.2 describes
the Ethernet Data Link.

X.25 Levels 2 and 3. Levels 2 and 3 operate over level 1 of the CCITT
recommendation X.25. This recommendation defines a standard interface
between data terminal equ1pment (DTESs) such as DECnet nodes, and a packet ~ switched data network. X.25 based data networks provide a virtual circuit service
between pairs of DTEs. Section 2.3 describes X.25 levels 2 and 3.

2.1 DDCMP Functional Description
- DDCMP
is a byte-oriented protocol. There are three ge'neral types of data link

\\‘\-_—«//‘

protocols: byte-oriented, character-oriented, and bit-oriented.
@A byte-oriented
protocol provides a count of the number of bytes that will be sent in the data portion of

each data message sent. In contrast, a character-oriented protocol uses special ASCII
characters to indicate the beginning of a message and the end of a block of text, while
a bit-oriented protocol uses flags to frame data, sent in undefined lengths.

DDCMP was designed in 1974 specifically for DNA. DDCMP is functionally similar

The Data Link Layer

2-1

to HDLC (High-level Data Link Control — the data link protocol adopted in 1975 by

the International Standards Orgainization), although HDLC is a bit-oriented
protocol. Another type of data link protocol, binary synchronous (BISYNC), i
character-oriented.

DDCMP is a general-purpose protocol: it operates on a variety of communication

systems from the very small to the very large. DDCMP makes maximum use-of
channel bandwidth and handles transparent data efficiently. (Data transparency is
the capability of receiving, without misinterpretation, data containing bit patterns
that resemble protocol control characters.
Character-oriented protocols cannot
handle transparent data as efficiently as byte- or bit-oriented protocols.) Other major
goals of the DDCMP design include error recording, to warn of impending channel
failure on degraded lines, and the provision of a simplified mode for bootstrapping and

testing functions.

DDCMP transmits data grouped into physical blocks known as data messages.
DDCMP provides a mechanism for exchanging error-free messages. A general
description of how this mechanism works follows:
|

DDCMP assigns a number to each data message beginning with number one (after
each initialization) and incremented by one (modulo 256) for each subsequent data
message. DDCMP places a 16-bit cyclic redundancy check (CRC 16) error detection
polynomial at the end of each data message transmitted.

The receiving DDCMP module checks for errors and, if there are none, returns the
message number with a positive acknowledgment of message receipt.
The receiving DDCMP need not acknowledge each message sent; acknowledgement
of data message n implies acknowledgment of all data messages sent up to and
including data message n.

If an error is detected by the receiving DDCMP, it uses time-outs and control
messages to resynchronize and trigger retransmission.

The principal DDCMP features are as follows:

Obtains data from the Physical Link layerin blocks con51st1ng of 8-bit bytes.

Sequences data by message numbers.

Pipelines up to 2565 messages. That is, it sends messages without waiting for
acknowledgment of each successive message.

Operates independently of channel bit width (serial or parallel) and transmission
characteristics (asynchronous or synchronous).

Operates witha wide variety of communication hardware and modems.
Detects errors by means of CRC-16 message trailers.

Retransmits to correct errors.
Achieves optimum performance with techniques such as pipelining, piggybacking
(that is, sending an acknowledgment within a returned data message), and
implying a positive acknowledgment
acknowledgment of current message.

2-2

messages

‘@

Operates in both half-duplex and full-duplex modes.

Supports point-to-point and multipoint communications.

Synchronizes transmission and reception on byte and message level.

The Data Link Layer

negative

Frames (envelops) data mesSages.
Provides a maintenance mode for diagnostic testing and bootstrapping functiOns.
Provides data transparency.

Notifies the other end of the link when restarting or initializing.
Maintains error counters.

Records the occurrence of events for automatic error reporting to the user.

'2.1.1 DDCMP Messages
There are three types of DDCMP messages:
- ®

Data (Section 2.1.1.1)

Control (Section 2.1.1.2)

Maintenance (Section 2.1.1.3)

Data messages send user data over
a physical link. Control messages return
acknowledgments and other control information. Maintenance messages consist of
the basic DDCMP envelope but contain information for down-line loading, up-line
dumping, link testing, or controlling a remote, adjacent system.

2.1.1.1 Data Messages
DDCMP formats all messages received from the Routing layer to be sent across the

physical link into a data message format (Figure 2-1).

The data message format

ensures proper handling and error checking of both the header information and the
data being sent. In the message format figuresin this chapter, the numbers below
each message field indicate its lengthin bits.

Data Message Format

SOH
COUNT

=
=

the numbered data message 1dent1fier
the byte count field

FLAGS
RESP
NUM
ADDR

=
=
=
=

the link flags
the response number
the transmit number
the station address field

BLKCK1

the block check on the numbered message header

BLKCK2

the block check on the data field

DATA

the n-byte data field, where 0<n=COUNT < ol4

Figure 2-1: DDCMP Data Message Format

The Data Link Layer

2-3

2.1.1.2 Control Messages
DDCMP has five control messages, Wh1oh' cai‘ry link control information,
transmission status, and initialization notification between DDCMP modules A brief
descr1pt1on of each message follows.

Acknowledge Message (ACK). This message acknowledges the receip‘t of correctly
numbered data messages that have passed the CRC-16 check.

The ACK message is

used when acknowledgments are required and when no numbered data messages are
to be sent in the reverse direction. The ACK message conveys the same 1nformat1on

as the RESP fieldin numbered data messages.

Negative Acknowledge Message (NAK).

The NAK message passes error

information from the DDCMP data-receiving module to the DDCMP data-sending
module.

The NAKTYPE field indicates the cause of the error.

The NAK message

serves two purposes:

It acknowledges receipt of all previously transmitted messages with a number
~ less than the current message number received (modulo 256).

‘@ It notifies the sender of error conditions related to the current message.

| Reply to Message Number (REP). The REP message requests received message

- status from the data receiver.
following conditions:
|

The data sender sends

a REP message under the

The data sender has sent a data message, and

The data sender has not received an acknowledgment of that data message, and

The time allocated for an acknowledament has explred

Start Message

(STRT).

The

STRT

message

establishes initial contact and

synchronization on a DDCMP link. The DDCMP module sends this message during
link start-up or re1n1t1a11zatlon

Start Acknowledge Message (STACK). The STACK message is the response to a
STRT message.
It tells the receiving DDCMP that the transmitting node has
completed initialization.

—

B F1gure 2-2 shows the formats of the DDCMP Control messages

2-4

The Data Link Layer

Acknowledge Message (ACK) Format

ENQ | ACKTYPE | ACKSUB| FLAGS | RESP
8

FILL

bl
7

ADDR | BLKCK3

Negative Acknowledge Message (NAK) Format

FILL | ADDR | BLKCKS3

Start Acknowledge Message (STACK) Format
ENQ |

STCKTYPE STCKSU%

NAKTYPE

REPTYPE

STRTTYPE
STCKTYPE

Mvrign””

REASON

REPSUB
STRTSUB
STCKSUB
FLAGS

RESP

FIIi[é

FIL[,‘,,D

ADDR

BLKCK3

the control message identifier
the ACK message type with a value of 1

ENQ
ACKTYPE

ACKSUB

FLAGS

the NAK message type with a value of 2
the REP message type with a value of 3
the STRT message type with a value of 6
the STACK message type with a value of 7
the ACK subtype with a value of 0
the NAK error reason
the REP subtype with a value of 0
the STRT subtype with a value of 0
the STACK subtype with a value of 0
the link flags
the response number used to acknowledge
received messages that checked out to be correct
a fill byte with a value of 0
the tributary address field

FILL

ADDR

BLKCK3

- the control message block check

Figure 2-2: DDCMP Control Message Formats
2.1.1.3 Maintenance Messages

The Maintenance Message, used in DDCMP maintenance mode, has the format
shown in Figure 2-3. This message is a DDCMP envelope for data controlling down-

line loading, up-line dumping, link testing, and controlling an unattended computer
system.

The DNA protocol for performing these functions is the Maintenance

Operation Protocol (MOP). MOP messages are sent within the DDCMP Maintenance
Message. See Chapter 7 for a description of MOP.

The Data Link Layer

2-5

/,.
3

Maintenance Message Format

| DLE

DLE

the maintenance message 1dent1fier

COUNT
FLAGS

=
=

the byte count field
the link flags

FILL

a fill byte with a value of 0

ADDR
BLKCK1

=
=

the tributary address field
the header block check on fields DLE
through ADDR

- DATA
BLKCK2

|BLKCK2
16

the n-byte data field, where 0 <n=COUNT < gld

the block check on the DATA field

" Figure 2-3: DDCMP Maintenance Message Format

212 DDCMP Operation
The DDCMP module has three functlonal components:
®

Frammg

Link management

Message exchange

Framing.

The framing component locates the beginning and end of a message
Framing involves locating a certain

received from a transmitting DDCMP module.

bit, byte, or message, and then recewmg at the same rate as subsequent bits, bytes, or
messages.

The modems and interfaces at the Physical Link level synchronize bits.
The

DDCMP

window

framing component synchronizes bytes by locating a certain 8-bit

the

bit

stream.

asynchronous

transmission techniques to synchronize bytes.

searches for a SYN character.
parallel links.

links

DDCMP

uses

start-stop

On synchronous links DDCMP

Byte synchronization is inherent in 8-bit multiple

DDCMP synchronizes messages by searching for one of the three

special starting bytes (SOH, ENQ, or DLE) after ach1ev1ng byte synchronization. To
maintain message synchronization, DDCMP:

Counts out fixed length headers.

When required, counts out variable length data based on the count field of the
header.

Link management.

The link management component controls transmission and

reception on links connected to two or more transmitters and/or receivers in a given

direction. Link Management controls the direction of data flow on half-duplex links
‘and the selection of tributary stations on multipoint links, using link flags.

addition, link management uses selective addressing to control the recelpt of data on
multipoint links.

2-6

The Data Link Layer

!
’

\v(
N
.

Message exchlange‘. The message exchange component'tran_sfers the data correctly

and in sequence over the link. Message exchange operates at the message level (after
framing is accomplished), exchanging data and control messages.

2.1.3.1 Typical Message Exchange
DDCMP is a positive-acknowledgment-with-retransmission protocol. This means that
for each data message correctly received and passed to the next level DDCMP returns
a positive acknowledgment.
Such acknowledgment is either an Acknowledge
Message (ACK) or a piggy-backed acknowledgment in the response (RESP) field of a
data message

If DDCMP receives a message out of sequence or with an error detected by the CRC,
DDCMP does not pass the data to the client. Typically, DDCMP does not acknowledge
this message.
Eventually a time-out occurs, and the data-sending DDCMP
retransmits the message. Alternatively, DDCMP may send a NAK to the datasending DDCMP module.
The transmission of DDCMP Data messages works as follows:

The transmitter increments the message number (modulo 256), putting it, n, in
the data message.

The message is transmitted within the required framing

envelope. A timer is started.

2. The receiver frames and receives the message, checks the received CRC value

againsta computed CRC value, and compares the message number with that
expected.
If the message checks out, the receiver returns a positive
acknowledgment (ACK) with that number, passes the message to the client, and
increments the next expected number to n + 1 (modulo 256). If the message does
not check out, the receiver ignores the message (or sends a NAK).

The transmitter then follows one of three procedures:
®

If the transmitter receives a positive acknowledgment, it checks the number

received to see if it is for an outstanding message.

If so, the transmitting

DDCMP notifies the client of successful transmission of that message as well
as of any previous, lower-numbered outstanding messages. If all outstanding
messages have been acknowledged, the timer is stopped. If one or more
messages remain outstanding, the timer is restarted.

If the transmitter receives nbthing, the timer expires. The transmitter sends
a REP message to initiate error recovery.

If the transmitter receives a negative acknowledgment, it retransmits the
message and all higher-numbered messages.

The transmitter may send several data messages before requiring acknowledgment of
the first one.
Acknowledgment of the highest numbered message implies
acknowledgment of lower-numbered messages. A negative acknowledgment implies

a positive acknowledgment of any previously transmitted lower-numbered messages.

Figure 2-4 shows a message exchange involving positive acknowledgment and
pipelining. Figure 2-5 shows error recovery from a NAK.
|

The Data Link Layer

2-7

NODE A

B
NODE

Messages

Client A requests

transmit of first data

@
DATA (message number n)

message

| Ack received
Client A requests

@ ‘Message received and
given to client B;

client B now expects

next data message to

ACK {of message number n)

ben+1

transmit of second
data message

Message received and

Message received

transmit of its
message. Acknowledgement of client A

client B requests

and client A requests

transmit. Acknowledgment of client
message B is piggybacked in RESP field
of data message.

DATA (message number m)

is piggymessage
backed in RESP field
of data message.

DATA (message number n +2)

Message received.

Client A requests

transmit

\-b

DATA (message number n + 3)

Message received.

Thése 3 data
messages are

pipelined.
Client A requests

transmit

Message received

DATA (message number m + 1)

@
ACK (of message number n + 1) @

Message received;

client B requests
transmit. Messages

n+2,n+3andn+4
are all acknowledged

with the acknowledgment of n + 4 in RESP

field of client B data
message.

ACK received

Figure 2-4: Typical DDCMP Message Exchange, Showing Positive

Acknowledgment, Piggybacking, and Pipelining

2-8

The Data Link Layér

NODEA
~Client
A requests

- transmit

o @

'NODE B

Messages
_

DATA (message number n)
B

NAK received

Retran.smlt |

Message received in
error

NAK (of message number n) | DATA (message number n)

Message received

Figure 2-5: DDCMP.Me'ssage Ex’change, Showing Error Rec'ovei'y a

2.1.3.2 Maintenance Mode
Maintenance mode uses the DDCMP framing and link management components, but
not the message exchange component. Sequencing or acknowledgment, if required,
must be handled within the data fields of the maintenance messages and is part of the
higher level protocol.

| The'Data Link Layer

.29

2.2 Ethernet Functional Description
The Ethernet local area network provides a communication facility for high speed
data exchange among computers and other digital devices located within a moderatesize geographic area. It includes a Physical layer and a Data Link layer. The primary
characteristics of the Physical layer are a data rate of 10 million bits per second, a
maximum station ‘separation of 2.5 kilometers, a maximum of 1024 stations, a

" shielded coaxial cable medium using base-band signalling, and support of a
branching, non-rooted tree topology. The primary characteristics of the Data Link
layer are a link control procedure using a fully distributed peer protocol with
~ statistical contention resolution and a message protocol Wthh supports variable size
frames and offers best-effort service.

The Ethernet is a local area network developed jointly by Digital Equipment
Corporation, Intel Corporation, and Xerox Corporation. The Ethernet specification
arose from an extensive collaborative effort of the three corporations and several
years of work at Xerox on a prototype Ethernet. The Ethernet specification provides a

'detalled spec1ficat10n of the two lowest layers of a network archltecture DNA Phase

The Ethernet Data Link layer defines a medium-independent link level
communication facility, built on the medium-dependent physical channel provided by
- the Ethernet Physical layer. It is applicable to a general class of local area broadcast
media suitable for use with the channel access discipline known as Carrier-Sense
Multiple-Access with Collision-Detection (CSMA/CD). The Ethernet Data Link layer
provides the following functions:
- ®

Data Encapsulation

- o
e

Framing (frame boundary delimitation)

Addressing (handling of source and destination addresses)
Error detection (detection of physical channel transmission errors)

Link Management

e Channel allocation (collision avoidance)
e Contention resolution (collision handling)
The split is reflected in the division of the Data Link layer into the Data Encapsulation sub-layér and the Link Management sub-layer, as shown in Figure 2-6,

2-10

The Data Link Layer

IV mcorporates the Ethernet into the PhysmalLink and Data Link layers and
“includes support of the Ethernetin the Routmg and Network Management layers.

Client Layer

Client-to-Data Linklnterface —f

Transmit

- Data Encapsulation

Transmit

Link Management

Receive

Data Decapsulation

Receive

Link Management

' *- Data Link-to-Physical Interface

[

Physical Layer

Flgure 2-6: Ethernet Data L1nk Layer Functions
The Ethernet Data Link prov1des a best-effort dehvery service. It does not provide an
error control facility to recover from transmission errors. In DNA the error control
functions necessary for reliable communications are provided by the End
Communication layer, using the Network Services Protocol (NSP). Additional error

control at the data link level is not required for Ethernet local networks, due to the
inherently low error rate of the Ethernet physical channel.

2.2.1 Ethernet Messages
The Ethernet Data Link has one type of message, called a frame.

The data

encapsulatmn function of the Data Link layer comprises the construction and
processing of frames. The sub-functions of framing, addressmg, and error detection
are reflectedin the frame format as follows:
®

Framing. No explicit frammg informationis needed since the necessary frammg

cues are presentin the interface to the Physical layer.

Addressing.
Two address fields are provided to identify the source and
|
|
destlnatlon stations for the frame.

Error Detection. A Frame Check Sequence (FCS) fieldis prov1ded for detection
|
of transm1ss1on errors.

Figure 2-7 below illustrates the Ethernet Frame format. In the figure the numbers
below each field indicate its lengthin bits.

The Data Link Layer

2-11

/
NI

DESTINATION|SOURCE | TYPE

TM\

Ethemet Data Link Frame Format
|DATA | FCS

= the destination data link address A data link address is

DESTINATION

one of three types:

‘o Physical Address: The umque address assoc1ated w1th
a particular station on the Ethernet. The 48 bit field
permits a station to have a unique address over all
a
Ethernets. In DNA Phase IV, each network node has
16 bit node address. The 48 bit Ethernet data link
address of a DNA Phase IV node is derived by prefixing
the 16 bit address with a 32 bit prefix assigned to DNA
Phase IV nodes. Thus, DNA Phase IV addresses are
unique over a single DNA network, which may include

multiple Ethernets, DDCMP links, and X.25 links.
e Multicast Address: A multi-destination address
associated by higher level convention with a group of
logically related stations on an Ethernet. DNA assigns
multicast addresses to the group of all Ethernet End
‘nodes and to the group of all Ethernet Routers (see
L

Section 3.2.1).

A distinguished, predefined
e Broadcast Address:
address which denotes the set of all stations on an
Ethernet.

= the source data link address. This field always contains the
physical address of the station transmitting a frame on the

SOURCE
o
o

TYPE

Ethernet.

= the type field. The type fieldiss reserved for use by higher
level protocols to identify the higher level protocol
‘associated with the frame, permitting multiple higher-level
protocols to coexist in the same Ethernet. Ethernet type

field values are assigned to the DNA Phase IV Routing
protocol and to the DNA Maintenance Operation protocols.
= the data field. The Ethernet data field contains higher level

DATA

e
FCS =~

protocol data andis 8n bits long, where 46 =n=1500. Full
transparency is provided, in the sense that any arbitrary
sequence of 8 bit bytes may appear in the data field. The
minimum length of the data field ensures that all frames
occupy the channel long enough for rehable collision

detection. .
= the frame check sequence This field contains the CRC-32

polynomial check on the rest of the frame.

- Figure 2-7: Ethernet Data Link Frame Format

DNA defines a standard convention for padding higher level protocols to the

“minimum Ethernet data field size. Higher level protocol data is preceded by a 16 bit
byte count and followed by any required pad bytes. When this convention is in effect,
Ethernet frames have the following format, and the COUNT DATA and PAD fields
have the meaning describedin Figure 2 8:

2-12

-The Data Link Layer

= the 16 bit count of DATA bytes, where COUNT=n

COUNT
DATA

PAD

= the data field. The Ethernet data field contains higher level
protocol data and is 8n bits long, where 0=n=1498. Full
transparency is provided, in the sense that any arbitrary
- sequence of 8 bit bytes may appear in the data field.
- = zero or more bytes of zeros, where m = max(0,44-n).

- Figure 2-8: Ethernet Frame Format with Padding

2.2.2 Ethernet Operation
This section prov1des an overview of the operation of the Ethernet Data Link layer
when transmitting or receiving frames. In this overview, the higher level user of the
Ethernetis referred to as the client layer. In DNA Phase IV Routing and Network

Management are clients of the Ethernet.

2.2.2.1 Transmission without Contention

When the Client layer requests the transmission of a frame, the Transmit Data
Encapsulation -component of the Data Link layer constructs the frame from the clientsupplied data, in the format specified by the client (with or without padding). The
Data Link layer appends a frame check sequence to provide for error detection and
hands the frame to the Transmit Link Management component for transmission.

Transmit L1nk,Management attempts to avoid contention with other traffic on the
Ethernet channel by monitoring the carrier sense signal provided by the Physical
Link layer, and deferring to passing traffic. When the channel is clear, frame

| ~ transmission is initiated. The Data Link layer provides a serial stream of bits to the
Physical layer for transmission.

The Physical layer transmits the frame, monitoring the medium and generating the

collision detect signal which, in the contention-free case under dlscussmn remains off

for the duration of the frame’s transmission.

When the frame has completed without contention, the Data Link layer so informs the
- Client layer and awaits the next request for frame transmission.
2.2.2.2 Reception without Contention
At the receiving station, the arrival of a frame is first detected by the Physical Link

- layer. As the encoded bits arrive from the medium they are decoded and passed to the
Data Link layer.

The Data Link Layer

2-13

Meanwhile, the Receive Link Management component of the Data Link layer, having
seen carrier sense go on, has been waiting for the incoming bits to be delivered.

Receive Link Management collects bits from the Physical layer as long as the carrier
sense signal remains on. When the carrier sense signal goes off, the frame is passed to
Receive Data Decapsulation for processing.

Receive Data Decapsulation checks the frame’s destination address field to decide
-

whether the frame should be received by the station. If so, the type fieldis checked to
ascertain which client module should process the incoming frame (for example

Routing or N etwork Management) If the client module specified that paddingis in
use for this protocol type, the frameis decapsulated appropriately. The contents of the
frame are then passed to the client with an appropriate status code. The status code is
generated by inspecting the frame check sequence to detect any damage to the frame

enroute and by checking for proper byte boundary alignment of the end of the frame.

2.2.2.3 Collisions: Handling of Contention
If multiple stations attempt to transmit at the same time, their transmitting data link
controllers may interfere with each others’ transmissions in spite of their attempts to
- avoid this by deferring. When two stations’ transmissions overlap, the resulting
- contention is called a collision. A station can experience a collision only during the

" initial part of its transmission (the collision window), before its transmitted signal
has had time to propagate to all parts of the Ethernet channel. Once the collision

window has passed, the station is said to have acquired the channel; subsequent
collisions are avoided, since all other stations can be assumed to have noticed the
signal, via carrier sense, and to be deferring to it.

In the event of a colhsmn the Physical Link layer first n0t1ces the 1nterference on the

~ channel and turns on the collision detect signal. This is noticed in turn by the
- Transmit Link Management component of the

Data

Link layer, and collision

handling beglns First, Transmit Link ‘Management enforces the collision by
transmitting a bit sequence called a jam. This ensures that the duration of the
collisionis sufficient to be noticed by the other transmitting station(s) involvedin the
,coll1s1on After the jam is sent, Transmit Link Management terminates the
transmission and schedules a retransmission attempt for a randomly selected time in
the near future ‘Retransmission is attempted repeatedly in the face of repeated
collisions. Since repeated collisions indicate a busy channel,
Transmit Link
Management attempts to adjust to the channel load by Voluntarfly delaying its own
retransmissions to reduce its load on the channel (backing off). Thisis accomplished
by expanding the interval from which the random retransmission time is selected on
each retransmission attempt. Eventually, either the transmission succeeds, or the
attempt is abandoned on the assumption that the channel has failed or has become
overloaded.

At the receiving end, the bits resulting from a collision are received and decoded by
the Physical layer just as are the bits of a valid frame. The Data Link’s Receive Link
Management component recognizes collision fragments ‘which are always shorter
than the shortest valid frame, and discards them.

2-14

The Data Link Layer

2.3 X.25 Functional Description
CCITT Recommendation X.25 defines a standard interface between an intelligent
system (Data Terminal Equipment or DTE) and an intelligent system that is an
~

access point to a public data network (Data Communications Equipment or DCE)
-operating in the packet mode. Public data networks offering the X.25 interface are

‘availablein many countries and represent an economical alternatlve to leased 11nes

and dial-up lines for many applications.

CCITT Recommendation X.25 is structured into three levels:

‘@

Level 1, the Physical level, defines‘t»he mechanical, electrieal, functional, and

procedural characteristics of the physical link between the DTE and the DCE. In
DNA, Level 1 resides in the Physical Link layer.

Level 2, the Frame level, defines the link access procedure for data exchange over
the hnk between the DTE and the DCE. In DNA Level 2 re51desin the Data Link
layer.

Level 3, the Packet level, defines the packet format and control procedures for the

exchange of packets containing control information and client data between the
DTE and the DCE. In DNA, Level 3 residesin the Data Link layer.

There‘ are two independent uses for X.25 in a Digital system or a DECnet network: (1)
as a Data Link module that allows two DECnet systems to communicate, and (2) as a

gateway to non-Digital systems accessible via a public data network. A single X.25
link can support multiple virtual circuits.. DNA Routing may use several virtual

circuits for communicating with DECnet nodes (see Section 3.3.4) while DNA X.25
Gateway Acess may use other virtual circuits for communicating with non-Digital
systems (see Section 6.3).

- 2.3.1 X.25Frame Level
The X.25 specification defines two alternative link access procedures (LAP and
-

LAPB) for Frame level communication between a DTE and a DCE. DNA supports the

LAPB procedure.

This procedure defines the control of full duplex communication

over a synchronous line between a DTE and a DCE.
functions:
'
®

LAPB performs the following

Provides an error free link by detecting transmlssmn errors and recovering from

such errors
®

Provides synchromzatlon to ensure that the Frame level entltles in the DTE and

DCE are in step

Detects procedural errOrs and reportsithem to the user of the _link.

The Data Link Layer

2-15

2.3.1.1 X.25 Frame Level Messages

Frame level modules in the DTE and DCE communicate by sending and receiving

~ frames to provide link control, data exchange flow control, and error control. There
are three types of frames

Information Frames (I Frames) Informatlon frames are used to transmit packet
level information and are also used to acknowledge prev1ous correctly received
Information frames.

Supervisory Frames (S Frames). Superv1sory frames are used to perform channel

control functions such as acknowledging correctly received Information frames,
requestlng retransmission of Informat1on frames and 1nd1cat1ng temporary inability
:

to receive Information frames

Unnumbered Frames (U Frames) Unnumbered frames are used to provide
| ad.d1t10na1 hnk control functions such as connectlng and d1sconnect1ng the link.
Frames are divided into commands and responses. Table 2- 1 summarizes the frame
types.

Table 2-:1:- X;25"Levvel 2 Frame’[‘ypes
G

B
S
o

(I)
Information Transfer

Responses

]

~_~_Commands_

Format

S
|

'
A
,

Receive Ready (RR)
Receive NOT Ready (RNR)

Reject(REJ)

‘Receive Ready (RR)
Receive NOT Ready (RNR)
'Reject (REJ)

Connect Link (SABM)

Disconnect (DISC)

Unnumbered Acknowledgment
(UA)

~_ Disconnected Mode (DM) -

Frame Reject (FRMR)

2.3.2 X.25 Packet Level
| The X.25 Packet level prov1des for multlple v1rtual connectlons between a DTE and
other DTEs connected to the public data network. A virtual connection between two
DTEsis called a virtualcircuit. Data transfer on a virtual circuit has the followmg
properties:

2-16

Itis full duplex. Packets may be transmittedin both directions simultaneously.
Itis sequential. Packets are delivered by the network to the remote DTE in the
same order as transmitted.

The Data Link Layer

® It is reliable. Packets are delivei‘ed without corruption of the Data field and are
not duplicated by the network. Packets may be lost, but loss will be reported by
- the network to the transmitting DTE (normally by resettmg or clearmg the
circuit).

A ’logical channel is an association between a DTE and its DCE for a given virtual
circuit. A logical channelis identified by a logical channel number. For each virtual
circuit a logical channel is assigned at both DTE/DCE interfaces. This number is
allocated independently at both DTE/DCE interfaces. Each DTE 1dent1fies a virtual

circuit by its local loglcal channel number.

Two types of virtual circuits are supported:

Switched Virtual Circuits (SVC). A switched virtual circuit is a temporary
association between two DTEs. In DNA, switched virtual circuits connecting DECnet
nodes are established and destroyed using Network Management functions. Switched
virtual circuits connecting DECnet nodes to non-D1g1ta1 systems are established by
-

users of X.25 Gateway Access.

Permanent Virtual Clrcults (PVC) A permanent virtual circuit is a virtual circuit
between two DTEs that is always established. A logical channel is permanently
allocated at each DTE/DCE interface to a permanent virtual circuit.

2.3.2.1 X.25 Packet Level Messages

The DTEs attached to a virtual circuit access the circuit by exCha.ngirig X.25 level 3

- packets with their local DCEs, using X.25 level 2 procedures. Table 2-2 summarizes
-

- the X.25 level 3 packet types and their functions.

The Data Link Layef |

2-17

‘Table 2-2: X.25 Level 3 Packet Types

Packet Type

DTE toDCE | Assigns a logical channel to the DCE and establishes a

Call Request

Description

) Dlrectlon

;

| Incoming Call

virtual circuit to the DTE addressed in the packet.

DCE to DTE | Indicates that a remote DTE wishes to establish a virtual

circuit with the receiving DTE and assigns a logical channel.

Call Accepted |

DTE toDCE | Indlcates that the DTE accepts the estabhshment of the

Call Connected

DCE to DTE | Indicates

‘|

virtual circuit.

that: the

remote

DTE

has accepted

the

establishment of the virtual circuit.

Clear Request

‘

DTE to DCE | Indicates that the DTE wishes to deassign the logical
channel and destroy the virtual circuit.

DCE Clear Co_nfirmatipn | DCE toDTE | Informs the DTE that the logical channel is deassigned.

DCE to DTE | Indicates tfiat the remote DTE destroyed the virtual circuit.

Clear Indication
DTE Clear Confirmation
‘

DTE to DCE | Informs the DCE that the virtual circuit has been destroyed
Co

and deassigns the logical channel.

DTE Data

DTE to DCE | Carries user data from the DTE over the logical channel.

DCE Data

'DCE toDTE | Carries data from the remote DTE over the logical channel

DCE RR

DCE to DTE | Updates the DTE transmit flow control information.

DCE RNR

DCE to DTE'| Indicates a temporary inability of the DCE to accept data

DTE RR
o

DTE to DCE | Updates the DCE flow control information, authonzmg the
'
- transmission of additional DCE Data packets.
'

DTE Interrupt

DTE toDCE | Carries user interrupt data over the logical channel.

packets. This condition is cleared by a DCE RR packet.

DCE Interrupt Confirmation | DCE to DTE | Acknowledges receipt of a DTE Interrupt packet.
DCE to DTE | Carries Interrupt data from the remote DTE.

DCE Interrupt

DTE Interrupt Confirmation | DTE to DCE | Acknowledges receipt ofa DCE Interrupt packet.

Reset Request

DTE toDCE | Requests that the logical channel and virtual circuit be set to

DCE Reset Confirmation

DCE toDTE | Acknowledges that the logical channel has been reset.

Reset Indication

DCE to DTE | Indicates that the logical channel hasbeen reset.

DTE Reset Confirmation

DTE to DCE

DTE Restart Request

DTE toDCE | Requests that all logical channels for SVCs be deassigned

an initial state.

Acknowledges the resetting of the logical channel.
andthatall PVCs be reset.

DCE Restart Confirmation | DCEtoDTE | Acknowledgesthe Restart Request

Restart Indication

DCE toDTE | Indicates that all logical channels for SVCs should be

DTE Restart Confirmation

DTE to DCE | Acknowledges a Restart Indication

Diagnostic

DCE to DTE

deassigned and all logical channels for PVCs should be reset.

Indicates to the DTE an error on one of its logical channels.

2-18

The Data Link Layer

2.3.2.2 X.25 Packet Level Operation
In response to a user request, the Packet level initiates the establishment of a virtual
circuit to the specified DTE. The Packet level assigns a logical channel number and

transmits a Call Request packet, specifying the logical channel number and the
address of the remote DTE.

- The remote DCE to which the DTE addressed in the Call Request packet is connected
assigns a logical channel number and transmits an Incoming Call packet over the

logical channel. The called DTE accepts the call by transmitting a Call Accepted
packet over the logical channel. In DNA, fields in the Incoming Call packet (such as
the Calling DTE Address, Called DTE Subaddress, and the Data field) determine
“which DNA Module handles an incoming call. (The called DTE can reject an
incoming call by transmitting a Clear Request packet.)

Once the virtual circuit has been established, data packets can be transferred. Each

data packet is sequentialy numbered (modulo 8). A Send Sequence Number is carried
in each data packet and specifies the position of the packet in the sequential stream.
A Receive Sequence Number is used to authorize the transmission of additional
packets (by acknowledging received packets) and is carried in Data Packets, Receive

Ready packets, and Receive Not Ready packets.

Flow Control
Flow control takes place separately for each direction of transfer on a logical channel,
making use of the packet sequence numbers. Flow control is based on the concept of a
window. - A window is a range of packets authorized to be transmitted across the
DTE/DCE interface. The window size is selected independently for each direction of
transfer on a logical channel. The range of packets in the window changes as packets
- are received and acknowledged by the destination DTE packet level.

Flow control can also utilize Receive Ready and Receive Not Ready packets. In DNA,

this method of flow control is employed when transmitting data to a DCE. Receive

Not Ready packets are never used when receiving data from a DCE.

Logical Messages
A DTE can construct a logical message from a number of consecutive packets, by
using the More Data bit in the Data packets. The More Data bit is set in each packet
except the last in a sequence of packets which compose a logical message. Data
packets also contain another control bit, the @ bit, which allows a transmitting DTE
to define two categories of data.

Interrupt Data
Each direction of transmission on a virtual circuit may have an outstanding Interrupt.
This interrupt flow allows a user to transmit
one byte of information not subject to the
rules of flow control which apply to normal data packets.
Interrupt data is
transmitted via an Interrupt packet. Interrupt datais acknowledged by an Interrupt
Confirmation packet.

The Data Link Layer

2-19

" Reset Procedure
A Reset procedure may be used to initialize a virtual circuit. Either DTE or the public
data network can initiate a reset. All data and interrupt packets in transit at the time
of a reset are discarded by the network. When the network wishes to indicate that the
virtual circuit is being reset, it transmits a Reset Indication packet containing a
reason for resetting. When a DTE wishes to reset a logical channel it transmits a

Reset Request packet. A reset of one logical channel for a virtual circuit will cause a
reset of the corresponding remote logical channel. A DCE confirms a Reset Request

packet by transmitting a DCE Reset Confirmation packet. A DTE confirms a Reset
Indlcatmn packet by transmlttmg a Reset Confirmation packet.

; »Clearing Procedure
Either DTE associated with a switched virtual circuit can destroy the circuit by
Clearing. The DTE transmits a Clear Request packet. The DCE returns a DCE Clear
Confirmation packet which deassigns the logical channel. The Clear Request is
reported
to the remote DTE in a Clear Indication packet. The remote DTE replies
with a DTE Clear Confirmatmn packet.

The loglcal channel at the remote DTE is

| then deass1gned

Restart Proc‘edUre |
The Restart procedure allows all switched virtual circuits associated with a DTE to be

Cleared and all permanent virtual circuits to be reset after a catastrophic failure. A
DTEissues a Restart Indication packet each time the DTE level 2 module connects to

the DCE. A DCE can transmlt a Restart Indlcatlon packet to force a DTE to clear all
| 1ts v1rtual circuits.

‘ Permanent Virtual Circuits (PVCs)

Permanent virtual circuits operate in the same manner as switched virtual circuits
except that no call setup is required. Legical channel numbers are permanently
~ assigned to PVCs. After a Restart, all PVCs are usable for data transfer. In DNA,
Network Managment defines the circuit name and parameters of each PVC known to
the packet level.

2-20

The Data Link Layer

User

Network Management

Network Application

Session Control

Data Link

Physical Link

Chapter 3

The Routing Layer
The Routing layer provides a message delivery service. The Routing layer accepts

‘messages (called packets within the context of Routing) from the End Communication
layer in a source node, and forwards the packets, possibly through intermediate
nodes, to a destination node. Routing implements a Datagram Service. A Datagram

service delivers packets on a best-effort basis. That is, the Routing layer makes no
absolute guarantees against packets being lost, duplicated, or delivered out of order.
Rather, a higher layer of DNA provides such guarantees (see Chapter 4, The End
Communication Layer).

The Rbuting layer selects routes based on network topology and operator-assigned

circuit costs and adapts to changes in the network topology. For example, Routing can
find an alternate path if a circuit or node fails. Routing does not adapt to traffic
loading. The amount of traffic on a circuit does not affect the path selected by Routmg

s
g pogn

for forwarding packets.

3.1 Routing Functional Description
- The Routing layer provides the following functions:

Determines packet paths A pathis the sequence of circuits and nodes) between
~ a source node and a destination node. If more than one path exists for a packet,
Routing determines the best path.

Forwards packets. If a packetis addressed to the local node, Routmg delivers it
to the End Communication layer. If a packet is addressed to a remote node,

Routing forwards it on the next adjacency in the path.

Adapts to topology changes. If a c1rcu1t or node falls on a path, Routmg finds
an alternate path, if one exists.

Adapts to different kinds of circuits. Routing supports paths consisting of
multiple types of Data Link circuits, including point-to-pomt and multi-point
lines, Ethernets, and X.25 virtual circuits.

)

Perlodlcally updates other Routing Modules. Routing modules
in other nodes

are periodically updated to be aware of any topology changes (such as c1rcu1ts
down or up).

The Routing Layer

3-1

Returns packets addressed to unreachable nodes. Routing returns packets

addressed to unreachable nodes to the End Communication layer, if requested to

do so by the End Communication layer.

- ®

Manages buffers.

Routing manages the buffers at nodes that are capable of

routing packets from a remote source to a remote destination.

®
|

Limits the number of nodes that a packet can visit.

Routing prevents old

packets from cluttering up the network.

- ®

Performs node verification. Routing will exchange verification passwords with
an adjacent node if so requested by Network Management.

Monitors errors detected by the Data Link layer.

Maintains counters and

gathers

event data for network

purposes.

management

Routing allows for a functional subset, permitting two types of Phase IV Routing
implementations:

Routing. Sometimes called full routing, this implementation type contains the
full complement of Routing components. A Phase IV routing node can:

Send packets from itself to any other Phase Iv node.»
‘Receive packets from any other Phase IV node.

® Route packets from other nodes through to other nodes. This is referred to as
route-through or packet switching.
2.

Nonrouting.

This

implementation

type

contains

subset

of the

Routing

components-' A'nonrouting node can:

Send packets from itself to any other Phase IV node.

Recelve packets addressed to itself from any other Phase IV node.

Nonrouting nodes can be placed only as end nodes in a network. An end node may be
connected to a network by only one active circuit. An end node does not require the
packet sw1tch1ng components Such a node, therefore, has less routing overhead.
Phase IV nodes are compatlble with Phase IIT nodes. Phase IV nodes differ from
Phase III nodesin the following ways:
®

Phase IV nodes canbe attached to an Ethernet.

L Phase IV nodes can handle larger networks than Phase I nodes. A Phase III

node cannot communicate with, nor be on the path to, nodes with addresses

out of the range that the Phase III node can handle.
e

Phase IV packet formats differ from Phase III in minor ways. Phase IV nodes

translate packet formats into Phase III format when communicating with a

Phase III node.

The logical distance from one node to another in a network is a hop. The complete
distance that a packet travels from source to destination is the path length. Path
length is measured in hops. The maximum number of hops the routing algorithm will

3-2

The Routing Layer

forward packets is called maximum vzszts This number is limited to 63 by the
Routlng architecture.
|
Routing allows the network manager to assign a cost to each circuit connected to a
- node. Cost is an arbitary integer (within the size limits of the cost field). Cost is used

in the routing algorithm that determines the best path for a packet.

The Routing

layer routes packets on the path of least cost, even if this is not the path with the

fewest hops.

The DNA Routing Functional Specification does not dictate how to

assign circuit costs, but it does suggest a cost assignment algorithm based on circuit

bandwidth. Both cost and maximum hops are values the system manager at each

node assigns using Network Management software (Chapter 7).

Figure 3-1 illustrates some of the Routing terms.

The glossary contains precise

definitions of these and other Routing terms (including network diameter, maximum
path length, maximum visits, maximum path cost, and maximum cost).

Legend:

G-

node

circuit

= circuit cost

- e

Node A w‘ants‘to send a packet to Node D. The‘re are three possible paths
PATH

PATHCOST

AtoB,BtoC,CtoD

®+®+@:7*

AtoB,BtoD

AtoB,BtoF,FtoE,EtoD

PATHLENGTH

3 hops

@+®=9

2 hops

@+®+@+@=13

4 hops

*7 is the lowest path cost; Node A therefore routes the packet to Node D via this path.
Figure 3-1: Routing Terms

The Rofiting Layer

3-3

/
\\.

Routing does not automatlcally adJust to traffic flow since packets always travel on
the path havmg the lowest cost.

The Routmg layer routes packets to nodes by usmg numerical addresses. Because

users often prefer to refer to nodes by name rather than by address, network node

names must be mapped to unique network node addresses. A nodeiis known to other
" nodes by one (optlonal) name and one address, both of which are unique in the

network.

However 1mp1ementatlons can optionally provide local users with the

capability of assigning additional node names that map to network-known names
and/or addresss. This process occurs at the Session Control layer (Chapter 5).

3.2 Routing Messages
There are two types of Routing messages: data packets and control messages. Data
packets carry data to and from the End Communication layer. Routing adds a packet
route header to messages sent by the End Communication layer. Control messages

exchange information between Routing modules in adjacent nodes to initialize the
Routing layer, maintain routing data, and monitor the states of adjacencies.

expansion.

Flgure 3-2 shows the Data packet routing header formats. The numbers under fields

in this and the following Routing message formats indicate the lengths of the fieldsin

bits.

3-4

The Routing Layer

There are two formats for data packet routing headers, depending on whether the
adjacent nodeis on an Ethernet or some other type of circuit. The Ethernet Endnode
Data Packet formatis used when communicating directly with an Ethernet end node
(a non-routing node directly connected to an Ethernet). The other format, the Phase
IV Data Packet format, is used in all other cases. The Ethernet Endnode Data Packet
Format has expanded addressing and reserved fields unusedin Phase IV, for later

| 3;.2.1 Packet Route Header

Phase IV Data Packet Routing Header Format
RTFLG

DSTNODE |SRCNODE [FORWARD

8
16
16
8
Ethernet Endnode Data Packet Routing Header Format
RTFLG | DSTNODE

|SRCNODE |

RES | FORWARD | RES

the set of flags used by the routing nodes, including:
o Return to sender flag (indicates whether or not the

RTFLG

packetis being returned)

e Return to sender request flag (mdlcates whether to
-

DSTNODE

SRCNODE
FORWARD
RES

=
=

discard or try to return packet)
|
o Intra-Ethernet Packet (indicates to the Ethernet
Endnode receiving this packet that the source of the
packet is on the same Ethernet, and can be
communicated with directly).
the destination node address

the source node address
|
the number of nodes this packet can visit
a reserved field

Figure 3-2: Packet Route Header Formats

- 3.2.2 Routing Control Messages

There are six types of Routing Control messages. One type, the Routing Message is

used on all types of circuits. Two types of Routing Control messagesare used only on
- Ethernets: the Ethernet Router Hello Message, and the Ethernet Endnode Hello
Message. The final three types of Routing Control messages, the Hello and Test
Message, the Initialization Message, and the Verification Message are used only on
non-Ethernet circuits. The following paragraphs discuss each of these messages.

Routing message.

The Routing | message provides information necessary for

updating the routing data base of an adjacent node. The information contained in the
Routing message is the path cost and path length for a set of destinations.

" Ethernet Router Hello message. The Ethernet Router Hello message is used for

initialization and periodic monitoring of Routers on an Ethernet circuit. Each
Ethernet router periodically broadcasts an Ethernet Router Hello message to all other
nodes (both Routers and Endnodes) on the same Ethernet circuit using a multicast
operation. The message contains a list of all Routers on the Ethernet circuit from
which the sending Router has recently received Ethernet Router Hello messages. By

exchanging Ethernet Router Hello messages, all Routers remain informed of the

status of the other Routers on the Ethernet circuit. The message also provides input
to an algorithm which selects a single Router on the Ethernet circuit to be the
Designated Router for that Ethernet. The designated Router assists two end nodesin

d1scover1ng that both are on the same Ethernet, by forwarding a packet from one to

The Routing Layer

the other as an Intra-Ethernet packet. Subsequent communications between the end
nodes can be d1rect

Ethernet Endnode Hello message. The Ethernet Endnode Hello message is used

for initialization and periodic monitoring of Endnodes on an Ethernet circuit. Each
Ethernet Endnode periodically broadcasts an Ethernet Endnode Hello rnessage to all
Routers on the Ethernet circuit using a multicast operation. The message is used by
Ethernet Routers to malntam the status (up or down) of Endnodes on the Ethernet.

Hello and Test message.

The Hello and Test message tests an adjacency to

‘determine if it is still operational (that both the circuit and the adjacent node are up).
Routing sends this message periodically on non-Ethernet circuits in the absence of
other traffic. Upon receipt of this or any other valid message, Routing starts (or
restarts) a timer. If the timer expires before another message is received from that
node, Routing considers the adjacency down.

Initialization message.

Routing sends this message when initializing a non-

Ethernet circuit. The message contains information about the node type, required
verification, maximum Data Link 1ayer receive block size, and Routing version.
Verification message. This message is used for verification purposes on nonEthernet circuits if the Initialization message indicates it is requlred
Figure 3-3 summarizes the Routing control message formats.

3-6

The Routing Layer

Routing Message Format
CTLFLG | SRCNODE | RTGINFO |CHECKSUM
16

16n

Ethernet Router Hello Message Format
CTLFLG

VERS

INFO

LIST

56m

Ethernet Endnode Hello Message Format
CTLFLG

VERS

INFO

168

Hello and Test Message Format
CTLFLG | SRCNODE
16

TEST DATA

8-128

Initialization Message Format
CTLFLG | SRCNODE | INITFO
16

Verification Message Format
SRCNODE | FCNVAL

CTLFLG

8-64

‘Routing control flag, with the following types :

CTLFLG

Initialization message

Verification message
Hello and Test message
Routing message
Ethernet Router Hello message

Ethernet Endnode Hello message
RTGINFO

Identification of source node’s Routing Module
Path length and path cost to a set of n destinations

- CHECKSUM
VERS

- One’s complement add check on routing information
Routing module version number

- SRCNODE
-~

Identification of node sending message

INFO

Node type (Router or Endnode), maximum Data Link
layer receive block size, hello timer
list of known Routers on the Ethernet circuit (Each of the

LIST

m entries consists of a 7 bit priority field for selecting the
Designated Router, 1 bit to indicate two-way connectivity,

and the 48 bit identification of a Router).

- TEST DATA
-

INITFO

FCNVAL

’

“sequence of up to 128 bytes of data to test the circuit
‘node type, required Verification message, maximum Data
Link layer receive block size, Routing version.
|

type-dependent verification information; function value.
Figure 3-3: Routing Control Message Formats

The Routing Layer

3.3 Routing Operation
The Routing layer consists of the following two sublayers with associated functional
components:

Routing Control Sublayer
®

Routing (Section 3.3.1)

Congestion control (Section 3.3.2)

]

Packet lifetime control (Section 3.3.3)

Routing Initialization Sublayer
®

Initialization (Section 3.3.4)

-3.3.1 Routing |
This component contains four processes described bélow::
®

Decision

Update

Forwarding

)

Recelve

Decision. The decision process selects routes to each destination in the network.

consists of a connectivity algorithm that maintains path lengths, and a traffic
assignment algorithm that maintains path costs.

When a routing node receives a

Routing message, the routing node executes the connectivity and traffic assignment

algorithms.

This execution results in updates to the data bases used to determine

packet routes.

Update.

The update process constructs and propagates Routing messages (see

Section 3.2.2).

The update process sends Routing messages to adjacent nodes as

required by the decision process and periodically to ensure the integrity of the routing

data bases.

Forwarding. The forwarding process supplies and manages the buffers necessary to
support packet route-through to all destinations. It performs a table look-up to select

the output circuit for the packet. If a destination is unreachable, this process either
returns the packet to the sender or discards it, depending on the option flagged in the

packet route header.
Receive. The receive process inspects a packet’s route header, dispatching the packet
to an appropriate Routing control component or to the End Communication layer.

3.3.2 Congestion Control :
Congestion control consists of a single process, transmit management. This process
manages buffers by limiting the maximum number of packets on a queue for a circuit.

If a queue for a particular circuit reaches a predetermined threshold, additional

3-8

The Routing Layer

packets for that queue are discarded to prevent congestion. Transmit management
also regulates the ratio of packets received directly from End Communication to route
-through packets. This regulatlon prevents locally generated packets from degrading
route- through service.

3.3.3 Packet Lifetime Control
Packet lifetime control prevents excessive packet looplng by dlscardlng packets that
have visited too many nodes

3.3.4 Initialization and Circuit Monitor

Routing initialization is a start-up procednre and periodic procedure for adjacent

nodes which ascertains neighbor node identities, and adapts to the characteristics of
the circuit which provides communication to that neighbor (such as Data Link block
size).

Initialization and monitoring operation are dependent on the type of circuit and the
type of adjacent node (a circuit and the node at the other end of the circuit are
collectively called an adjacency). The following paragraphs discuss these operations
for the types of circuits supported by the Phase IV Rout1ng Layer

3.3.4.1 Ethernet Circuits

On an Ethernet circuit, initialiiation and monitoring are perforrned by the Ethernet

Endnode Hello message and the Ethernet Router Hello message. The initialization
and monltorlng funct1ons of these messages are descrlbedin Section 3.2.2.

3.3.4.2 DDCMP Clrcmts
On DDCMP circuits (both point-to-point and multi-point), Initialization and

monitoring are performed by the Initialization message, the Verification message,
and the Hello and Test message. In addition to the functions describedin Section 3.2.2
for these messages, the Initialization message adapts the Routing layer’s operation to
the block size of the physical line as managed by the Data Link Layer.

3.3.4.3 X.25 Circuits.
On X.25 virtual circuits (both Permanent and Switched) the Initialization and
Verification messages are used as for DDCMP circuits, except that the Data Link
block size is taken from the X.25 Data packet size selected when the virtual circuit is
initialized.

X.25 circuits are initialized when the Routing layer is so commanded by Network
Management. The Routing Initialization sub-layer contains a database, maintained
by Network Management, which consists of a mapping between Routing layer circuits
and X.25 Packet level Permanent Virtual Circuit identifiers in the case of Permanent
Virtual Circuits, and between Routing layer circuits and virtual call parameters in
the case of Switched Virtual Circuits. The database contains items such as DTE

‘subaddress range, maximum packet size, number of times to reattempt failed calls,
ete.

When placing a virtual call, the Routing Initialization sub-layer provides the
necessary parameters to the X.25 Packet level component of the Data Link layer,
which places the call to the remote Routing layer at the supplied foreign DTE address.

The Routing Layer

3-9

If the call is accepted (see below) Routing In1t1a11zatlon proceeds with olrcult'
1n1t1a11zat1on as descrlbed for DDCMP.
The Routing Initialization layer listens forincoming virtual calls on circuits managed

by the X.25 Packet level component of the Data Link layer.

When informed of an

incoming call by the X.25 Packet level, Routing Initialization executes an algorithm

to determine whether to accept the call, reject the call, or leave the call pending in
case some other module of DNA

(such as the X.25 Gateway Server module - see

Section 6.3.2) wishes to process the call. The algorithm uses the mapping database
described above and makes its decision based on comparing the foreign DTE
subaddress in the Incoming Call packet to the DTE subaddress range in its database.
If the decision is to accept the call, 1n1t1a11zatlon proceeds as descrlbed in the
preceding paragraphs.

Once an X.25 circuit is established, the Routing Initialization sub-layer performs |
some additional functions not performed for Ethernet or DDCMP circuits, in order to
maximize the performance and 1ntegr1ty of X.25 circuits.

These functlons are as

~ follows:
®

Blocks and deblocks Routing layer messages.

Since communication over

X.25 circuits is in the form of (usually small) packets, Routing Initialization will
block and deblock DNA Datagrams into X.25 Packetsin order to minimize the
number of packets transmitted over the virtual circuit.
®

Fragments and Reassembles Routing layer messages.

If the X.25 virtual

circuit packet size is too small to contain an entire Routing layer message,
Routing Initialization will fragment and reassemble Routing layer messages so
that they will fit in the available X.25 packet size.

Checksums X.25 packets.

‘

The Routing Initialization sub-layer appends a

checksum to every Data packet transmitted to insure the integrity of data
transmitted over the virtual circuit.

As with other types. of circuits, the Routing Initialization sub-layer monitors X.25
circuits for non-recoverable errors. These include X.25 Resets, Restarts, and Clears.
If any of these events happen, Routing Initialization declares the circuit down, S0

informs the Routing Control sublayer, and attempts to re-initialize the circuit.

Figure 3-4 summarizes the operation of the Routing layer.

3-10

The Routing Layer

receive interface

transmit
interface to End
Communication

to End
Communication

End Communication Layer

Decision Process

maintains

.
Contains:

let-—,

maintains

Routing Layer

Database

® selects paths

Contains: _
r

reachability
vector

output

circuit to be

Database

Routi

D:tlagaalses outing

-« -9

formation

Database

(hop, cost,

flags, etc.)

used

v ¥

Routing In-

Forwarding

gs.

ete.

Forwarding Process
e

supplies and manages buffers

rejects packets from End
Communication if queue is too

Receive Process

Update Process

reads Forwarding Database

® receives packéts
® passes them to

® generatesand
propagates routing

full
®

and decides on which circuit to
send packet

takescare of packets for

dropsincoming

messages

appropriate process

-€

unreachable nodes

queue is already

for route-through

packets if the

discards packets
that have visited
too many nodes

(Loop Detection)

full

Control sublayer
Initialization sublayer§

DDCMP Initialization
® sends andreceives
Initialization,

Verification, and Hello
and Test messages

Ethernet Initialization X.25 Initialization

® sends andreceives

Ethernet Router and

Endnode H_ello messages
»o

selects Designated

Router

Data Link Layer

DDCMP Module

® initializes PVCs
®

issues outgoing and

| Ethernet Module

Database

accepts incoming calls

Contains:

blocks, deblocks,

fragments, and

reassembles packets

m
Circuit

mapping

X.25 Level 3 Module

and their Functions
Figure 3-4: Routing Layer Components

The Routing Layer

3-11

\"'«.—-‘/

User

Network Management

Network Application

Session Control

Routing

Data Link

Physical Link

Chapter4

The End Communication Layer
The End Communication layer, residing immediately above the Routing layer,

provides a system-independent process-to-process communication service that allows

two processes to exchange data reliably and sequentially, regardless of their locations

in a network. The Network Services Protocol (NSP) is the protocol of the End
Communication layer. Communication is on a connection basis. A connection
between two processes is called a logical link. A logical link permits two-way
simultaneous transmission of normal data messages and independent two-way

simultaneous transmission of interrupt messages.

4.1 NSP Functional Description
NSP perfo’rm's the following functions:
o Creates‘and destroys logical links.
®

Guarantees the delivery of data and control messages in sequence to a specified
|

destination by means of an error control mechanism.

and normal data from transmit buffers to
Manages the movement of interrupt

Breaks up normal data messages into segments that can be transmitted

receive buffers, using flow control mechanisms. -

individually, and reassembles these segments in proper sequence upon reception.

4.2 NSP Messages
NSP modules provide logical link service, flow control, error control, and other
functions by sending and receiving NSP messages. There are three types of NSP
messages:

Data

Acknowledgment

(Control

Table 4-1 summarizes the functions performed by each NSP message.

The End Communication Layer

4-1

Table 4-1:

NSP Messages

Type

Data

DeScription |

Message

Data Segment

Carries a portion of a Session Control
message. (This has been passed to Session

Control

from

higher DNA

layers

and

Session Control has added its own control
information, if any.)

Data ( also called

Carries urgent data, originating from

Interrupt

QOther Data)

higher DNA layers. [t also may contain an

optional Data Segment acknowledgment.
Carries data flow control information and

Data Request

optionally a Data Segment acknowledgment (also called Link Service message).

Interrupt Requést

| Carries interrupt flow control information

and optionally a Data Segment acknowledgment (Link Service message).

Acknowledgment

Data Aéknowledgment

Acknowledges receipt of either a Connect

Confirm message or one or more Data
messages, and optionally ‘an

Segment

Other Data message.

Othef Data

Acknowledgment

)

Acknowledges receipt of one or more
Interrupt, Data Request or Interrupt

Request messages, and optionally a Data

Segment message.
Connect

A‘ckno'wledges receipt of a Connect
- Initiate message.

Control

Connect Initiate and

Retransmitted

Carries a logical link connect request

from a Session Control module.

Connect Initiate

Connect Confirm

Carries logical link connect acceptance
from a Session Control Module.

Distonnec_t Initiate

Carries a logical link connect rejection or

disconnect request from a Session Control
module.

No Resources

Sent when a Connect Initiate message is
received and there are no resources to

establish a new logical link (also called
Disconnect Confirm message).

Disconnect Complete

Acknowledges the receipt of a Disconnect
Initiate message (also called Disconnect

Confirm message).=
No Link

Sent when a message is received for a
nonexistent

logical

link

- Disconnect Confirm message).
No Operation

4-2

The End Communication Layer

Does nothing.

(also

called

4. 3 NSP Operatlon

- NSP operatlon consists of four functions: |
®

Creation, maintenance, and destruction of loglcal llnks (Section 4.3.1)

Segmentation and reassembly of data (Section 4.3.2)

®.

Error control (Section 4.3.3)

Flow control (Section 4.3.4)

4.3.1 Logical Links
The primary functions of NSP are to create, operate, and destroy logical links at the
request of the Session Control layer. A logical link may be thought of as a full-duplex
logical channel between two users. The users are guaranteed that, in the absence of a
network failure disconnecting them, data sent on a logical link by one user will be
‘received by the other user in the order sent. An equivalent term for logical link thatis
often usedis virtual circuit. A logical link enables a user process at one end of the link
to send data to a user process at the other end of the link. The NSP mechanisms that
set up a link, check data for errors, and manage data flow are transparent to the user
processes.
|
|

There can be s-everal"iogiéai' links at ahy time, even between the same two NSP
implementations. Any Phase III or [V node can establish a logical link with any other

Phase III or IV node in the same network. Figure 4-1 illustrates typical connections.

| L Physical
Link

“\M

Figure 4-1: Logical Links

The End Communication Layer

4-3

NSP establishes, maintains, and destroys logical links by exchanging control
messages with other NSP modules or with itself. Figure 4-2 shows a typical message
exchange. In this figure, an NSP module first initiates a connection, then sends data,

and finally, disconnects the link, based on commands from Session Control. The
|

messages that control data flow are not shown.

NODE A

NODE B

UseratNode A requests connection.

CONNECT INITIATE J—-——>
® User at Node B accepts connection. -

User at Node A requests that a message (thatis two segments long) be transmltted NSP at Node

A sends the first data segm ent.

|—

NSP at Node B acknowledges receipt of the first data segment.

User at Node A sends the second data segment.

DATA

[
®

" _4—.——[CONNECT CONFIRM ]

—

NSP at Node B acknowledges recenpt of the second data segment and gives the message to the
user at Node B

@ User at Node A'r'eque‘sts a disconnection and NSP at Node' A sends a Disconnect Initiate message.

DISCONNECT |N|T|ATEJ—-—>
NSP at Node B receives Disconnect Initiate message sends a Dlsconnect Conflrm Message and
mforms the user at Node B.

*———f DISCONNECT CONFIRM ]

Figure 4-2: Typical Message Exchange Between Two Implementations of NSP

4-4

The End Communication Layer

/‘ g
N

NSP operates concurrentlyin both directions on a logical link (full-duplex). The user
process at either end of the l1nk can initiatea disconnection at any time.

You can think of a logical lmk as made up of two separate data subchannels, each
carrymg messages in both d1rectlons

Normal data subchannel.

between two NSP modules.

This subchannel carries Data Segment messages
~

Other Data Subchannel. This subchannel carries:
®

Interrupt messages

Data Request messages

Interrupt Request messages

4 3 2 Segmentatmn and Reassembly of Data
The Routing layer limits the amount of user data that NSP can sendin one datagram
Taking normal data from Session Control buffers, NSP breaks it up, if necessary, into

segments. Each segment is numbered and sent along with its number and other
control informationin a Data Segment message to the receiving NSP module. The
receiving NSP module uses the sequence numbers to reassemble the data segmentsin

correct sequence in the receiving Session Control buffers.
NSP segments normal data only: interrupts are not segmented because interrupt
datais limited in s1ze and w1ll always fitina smgle datagram

4.3.3 Error Control

, The NSPs at each end of a loglcal link pos1t1vely acknowledge received data.

Optionally, an NSP may negatively acknowledge received normal data that it must
“discard (for example, if the datais received too far out of order). If the transmitting
NSP receives a negative acknowledgment or fails to receive a pos1t1ve
acknowledgment durmg a t1meout interval, it retransmits the data.

' Figure 4-3 descr1bes data segmentatlon and acknowledgment.

'The End Communication Layer

4-5

The data-transmittlng NSP assigns a transmut num ber to the message, and starts a tlmer
transmlt number= n

"""
Data-transmitting

Data Segment Message

NSP

’

b —————

Data-receiving

NSP

transmit number = m
Other DataMessage

}~----—>»

Data Subchannels

®
®

If the timer times out, the message is retransmitted.
If the timer does not time out, and the flow control mechanism allows another message to be sent, the data-

transmitting NSP assigns the transmit number plus one to the next data message transmittedin that
subchannel.
transmitnumber = n + 1

T
Data-transmitting

Data Segment Message

B
)

transmitnumber = m + 1

—————

Other Data Message

-——=

Data-receiving

Data Subchannels

When the message wnth the flrst transmlt numberis recelved by the datarecelvmg NSP, it returns that
number as an acknowledgment number within the first acknowledgment.

® If the next data message transmit number received is equal to the current acknowledgment number plus one,
the data-receiving NSP accepts the data message, incrementing the acknowledgment number. It then sends
the new receive acknowledgment number back to the data-transmlttlng NSP within an acknowledgment
message.

Data-transmitting

NSP

<€

—————-——--=--17

—

receive ack.number = n 1 | receive ack. number =n+1

Data Acknowledgment ]
Message”

TTTTTTTTTTT Tt

Data Acknowledgment

Message "

Data-receiving

NSP

receive ack. number = m

¢ - — - - -~ —-—4

Other Data Acknowledgm ent

-----------

Message

Data Subchannels

*The data-receiving NSP might not send an acknowledgment for each data message received. The receive
acknowledgment number implies that all previous numbers were received.

However, if the data-receiving NSP receives a data message transmit number less than or equal to the current

receive acknowledgment number for that subchannel, the data segment is discarded. The data-receiving NSP
sends an acknowledgment back to the data-transmnttmg NSP. The acknowledgment contains the receive
acknowledgment number.

@ If the data-receiving NSP receives a data message transmit number greater than the current receive

acknowledgment number plus one for that subchannel, the data segment may be held until the preceding

segments are received or it may be discarded.

Figure 4-3: Acknowledgment Operation

4-6

The End Communication Layer

4.3.4 Flow Control

NSP’s flow control mechanisms ensure that data is not lost for lack of buffering

capability and that deadlocks do not occur. Both normal and interrupt data are flowcontrolled.

The data-receiving part of NSP controls data flow. When a logical link is formed,
each NSP informs the other of the way it wants to control the flow of normal data as a
data receiver. The receiving NSP chooses one of three types of normal data flow

control:
®

None.

_'Seg'm’ent." The receiver sends a request count of the number of segments it can
accept. See Figure 4-4.

- ®

Message. The receiver sends a request count of the number of Session Control
messages it can accept.

Note, message flow control is obsolescent and will be
|

eliminated at a future time.

These schemes define the use of a request count thatiis set by the receiver and used by
the transmitter to determine when data may be sent. In addition, the receiver can
always tell the transmitter either to stop sending data unconditionally or to start
sending data under the normal request count conditions. The receiver also controls
interrupt data flow with an interrupt request count.

The Data Request Message, Interrupt Message, Interrupt Request Message, and the

Other-Data-Ack Message may contain an acknowledgment for a Data-Segment
Message.
In addition, the Data-Segment Message and Data-Acknowledgment
Message may contain an acknowledgment for an Other-Data Message. This ability,
to combine acknowledgments with data and control messages, reduces the number of
NSP messages required per user message.

- Figure 4-4 illustrates

NSP Flow control operation:

The End Communication Layer

4-7

NSP/Node A
Data-Transmitting

NSP/Node B
Data-Receiving

NSP/Node A sends a Connect lnitiate to NSP/Node B:

Connect

Initiate

NSP/Node B, having received the Connect Initiate message returns a Connect Confirm message A fieldin the
message indicates that NSP/Node B expects segment flow control:
“Iwant segment flow control”~— TM\
)

Connect

Confirm

|
l

NSP/Node A's data base has the initial value of 0 for its request count variable for normal data segment flow
control:

FLOWrem
dat = 0 |
NSP/Node B sends a Data Request message contammg a flow control value that indicates the number of Data
Segment messages NSP/Node B can receive. (NSP/Node B executes an |m plementatlondependent algorithm to
determine this value):

“| can receive nsegments”
-

Data

-«

Request

—~—"\
'

|
1

NSP/Node A increments FLOWrem _dat by n:

FLOWrem
dat = n

NSP/Node A executes algorithm to determine if Data Segment num ber 1 can be sent (highest acknowledged

Data Segment message plus the current request count must be greater than or equal to the number of the next
Data Segment message sent):

0+FLOWrem_dat=1?

—NO —can’t send

YES -send

The answer to the above is YES, so NSP/Node A sends Data Segment number 1:
Data

Segment 1

.
o

NSP/Node B acknowledges receipt of first data segment:

Data Ackndwledgment
of segment number 1

® NSP/Node A subtracts 1 from its flow control request count:
[FLOWrem dat = n-1

Figure 4-4: Segment Flow Control Shown in One Direction on a Logical Link

4-8

The End Communication Layer

User

Network Management

New Application
End Communication
Routing

- Data Link
Physical Link

5
Chapter

“The Session Control Layer
The Session Control layer, residing immediately 'aboVe ‘the End Communication
layer, provides system-dependent, process-to-process ‘communication functions.
These functions bridge the gap between the End Communication layer and the logical

link functions required by processes running under an operating system.

5.1 Session Control Functional Descriplti'o_n
Session Control functions include:

‘Mapping nbdé names to node addresses. A Session Control module maintains

a node name mapping table that defines the correspondence between a node name
and either a node address or an adjacency. (An adjacency is used for loopback
testing under the control of N etwork Management.) The table enables the Session
Control module to select the destination node address or channel number for
outgoing connect requests to the End Communication Layer. For incoming
connect requests from the End Communication Layer, the Session Control module
uses the table to identify the node from which the request originated.
Identifying end users. A Session Control module executes a system dependent
algorithm to determine if an existing end user process corresponds to the
d an incoming connect request. It also
destination end user process specifiein

an existing
performs additional functions related to passinga connect requetost
|

end user process.

Activating or creating processes. A Session Control module may create a new

_process or activate an existing process to handle an incoming connect request.

Validating incoming connect requests. A Session Control module uses access

control information included in a incoming connect request to perform systemdependent validation functions.

The Session Control Layer

5-1

Session Control defines two data bases that are not implementation-specific:
®

A node-name mapping data base.

A data base conta1n1ng the states of Session Control and optmnal default
connection timers.

Figure 5-1 shows a model of Session Control operating within a net»wfork node.

A NETWORK NODE

)

- COMMENTS
-

END‘

USER
R

USER2|

PROCESSES

| enp

END

SYSTEM

DEPENDENT | |OPERATING |
SYSTEM

Communication layer for end user
it

® and ® are data bases used by Session
Control.

S\L(t);et:\s:soper;tnnugn:;ls:e‘; " coniunction

[-® |

‘

COMMUNICATION |

MODULES

|-®

- END

/
-

- Session Contrnl is en interface to the End

CONTROL

‘

SYSTEM

| SESSION

INDEPENDENT

USER 3

\ \ /

_—

USER 1 N

MODULES

End users are User, Network Appllcatlon,

~and Network Management” modules

The End Communication layer provides
the logical link service to Session Control.

functions

are

not

dependent

individual operating system:s.

LAYER

*
|
1

- THE NETWORK

Network Management interfaces with
Session Controlin two ways: (1) to obtain
access to the logical link service and (2) to

monitor

and

operations.

control

Session

Control

Figure 5-1: A Session Control'Model‘

5.2 Session Control Messages
Session Control’s message protocol defines messages sent on a loglcal link as connect
data, reject data and disconnect data.
»|
Figure 5-2 shows the formats for the Session Control messages. 'The numbers below
each message field 1nd1cate its max1mum lengthin bytes

5-2

The Session Control Layer

Connect Data Message Format

DSTNAME|SRCNAME |[MENUVER R'QSTRID PASSWRD [ACCOUNT (USRDATA
19

Reject/Disconnect Data Message Format
REASON | DATACTL
1

)

DSTNAME

SRCNAME
MENUVER
RQSTRID

=
=
=

ACCOUNT

PASSWRD

USRDATA
-REASON

=
=

DATACTL

=
|

the destination end user name

thesourceend user name
the field format and version format
o
the source user identification for access verification

the access verification password

the link or service account data
the end user process connect data
areasoncode
o

user data (length of field determined by the total length of
reject or disconnect data received from the End Communi-

cation layer)

Figure 5-2: Session Control Message Formats

)

5.3 Session Control Operation
Sessmn Control performs the following bas1c operatlons
®

Requests log1cal links on behalf of end user processes (Sectlon 5. 3 1).

Receives connect requests addressed to end user processes (Section 5.3.2).

Sends and receives logical link data (Section 5.3.3).

‘Disconnects and aborts logical links (Section 5.3.4).
Optionally, monitors logical links (Section 5.3.5).

5.3.1 Requestmg a Connection

Upon receipt of a loglcal link connect request from an end user process, Session
Control performs four tasks:
Identifies the destination node address or channel number for the End Communication layer by using the node name mapping table.
Formats connect data for the End Communication layer.
Issues a connect request to the End Communication layer.

Optionally starts an outgeing connection timer. Expiration of the timer prior to
an acceptance or rejection from the End Communication layer causes Session
Control to disconnect the logical link for the source end user process.

The Session Control Layer

5-3

5.3.2 Receivmg a Connect Request
Upon detection of an incoming connect request from the End Communlcatlon layer
:
Session Control performs six tasks:
®

Parses connect data to obtain such 1nformat10n as source and destmatlon end user

process names and access control information.

Validates any access control information.

Identifies, creates, or activates a destination end user process.
Maps the source node’s address or channel number to a node name, if there is one.
Delivers the incoming connect request to the end user process.

Optionally, starts an incoming timer when the connect request is delivered.

Expiration of the timer before the end user process accepts the connect request
causes Session Control toissue a reject to the End Commumcatmn layer.

5.3.3 Sending and Receiving Data
Thisis a system-dependent function that handles end user process requests to send

and receive data. Basically, the functions are passed directly to the End Communication layer.

5 3.4 Dlsconnectmg and Abortmg a Logical Link
Asin the case of sending and receiving data, Session Control passes end user process

disconnect and abort requests directly to the End Communication layer.

Similarly,

notification of a logical link dlsconnect or abort 1s passed directly to the end user
-

process.

5.3.5 Monitoring a Logical Link |
This is an optional system dependent function that may be used for the following
purposes:

Detecting probable network dlsconnectlon between the nodes at either end of the
logical 11nk

Detecting a failure, by the End Communication layer, to dellver transmitted data

in a timely manner.

5-4

The Session Control Layer

‘User
Network Management

Session Control
~ End Communication

Routing
Data Link

Physical Link

Chapter 6
The Network Application Layer

The Network Application Layer contains a number of independent, commonly used
modules. These modules access data or provide communication services to users.

Currently, five Phase IV DIGITAL- supplied protocols are specified for this layer:

The Data Access Protocol (DAP) This protocol permlts remote file access and
file transferin a manner thatis independent of the [/O structure of the operating
system beingaccessed. Section 6.1 describes DAP.

The Network Virtual Terminal Protocols. These protocols permit terminals

connected locally to a host DECnet system or to a Terminal Concentrator system
to access remote DECnet host systemsin order to issue interactive commands and
run apphcatlon programs Section 6.2 describes Network Virtual Terminals.
The X.25 Gateway Access Protocol.

This protocol permits user--written

modulesin a host DECnet system to commumcate with peer modules in a non-

DECnet system aeross an X.25- based Public Packet Switching Network. The user

module does not have to reside in the same DECnet system that is directly

connected to the X.25 Network (the DTE- Data Termmal Equlpment) Section
6.3 describes X.25 Gateway Access.

The SNA Gateway Access Protocol.
This protocol permits user-written
modules (and a number of DIGITAL-supplied modules)in a host DECnet system
to communicate with IBM host application programs and apphcatlon subsystems
in a Systems Network Architecture (SNA) network. The user modules do not

“have to reside in the same DECnet system that is directly connected to the SNA
network (the Gateway system). Section 6.4 describes SNA Gateway Access.
The Loopback Mirror Protocol. This protocol consists of one message used
between the Network Management loopback access routines and loopback mirror.
These modules test logical links. Section 7.2.3 describes this operation.

The Network Application Layer

6-1

6.1

DAP Functional Description

DAP provides the following functions and features:
| | Supports heterogeneous file systems.
|
|
|
Retrieves a file from an input device (a disk file, a card reader a terminal etc.).
| Stores a file on an output device (a magtape, a line printer, a terminal, etc.).
| Transfers files between nodes.

Supports deletion and renaming of remote files.
Llsts directories of remote files.

Recovers from transient errors and reports fatal errors to the user.
Allows mu1t1ple data streams to be sent over a logical link.

Submits and executes remote command files.

Permits sequential, random, and indexed (ISAM) access of records

Supports sequential, relative, and indexed file orgamzatlons

Supports wildcard file spec1ficatlon for sequent1al file retrleval flle deletion, file
renaming, and command file executmn
»
Permits an optional file checksum to ensure file 1ntegr1ty

DAP is designed to minimize protocol oVerhe‘ad; For example, defaults are specified
for fields wherever possible. In addition, a file transfer mode eliminates the need for
DAP control messages once file data flow begins. F1nally, relatively small file records
can be blocked together and sent as one large message.

DAP is a set of messages and the rules governing their exchange between two
- cooperating processes.
Section 6.1.1 'describes' DAP messages.
Section 6.1.2
- summarizes the operation of the cooperating DAP-speaking processes, ‘which provide
DECnet remote file access.

Sectlon 6.1.3 descrlbes the most common DAP-speaking

DECnet facilities.

6.1.1

DAP Messages

DAP-speaking processes-use the messages listed in Table 6-1 to accomplish remote
file access and transfer.

6-2

The Network Application Layer

TABLE 6-1:

DAP Messages

Function

Message

Exchanges system capability and configuration information

Configuration

between DAP-speaking processes. Sent immediately after a
logical link is established, this message contains information
about the operating system, the file system, protocol version,
|

and buffering capability.

~ Attributes

Provides information on how data is structured in the file

being accessed. The message contains information on file
organization, data type, format, record attributes, record

length, size, and device characteristics.

Specifies the file name and type of access requested.

Access

Continue-Transfer

- Allows recovery from errors. Used for retry, skip, and abdrt

Acknowledge

datastreams.

Sends Control information to a file system and establishes

- Control

after an error is reported.

Acknowledges access commands and Control messages used
|

to establish data streams.

Denotes termination of access.

Access Complete

Data

Transfers file data over the logical link

‘

Returns status and information on error conditions

- Status

Key Definition

Specifies key definitions for indexed files.

Allocation Attributes
Extension

Attributes Extension

Specifies the character of the allocation when creating or
explicitly extending a file.

Summary Attributes

Returns summary information about a file.

Date and Time Attributes

Specifies time-related information about a file.

- Extension

Extension

Specifies file protection codes.

Protection Attributes
Extension

‘Name

Sends name information when renaming a file or obtaining
file directory data.

6.1.2 DAP Operation
Two cooperating processes exchange DAP messages: the user process and the server
process that acts. on the user’s behalf at the remote node. User /O commands
accessing a remote file are mapped into equivalent DAP messages and transmitted
via a logical link to the server at the remote node. The server interprets the DAP
commands and actually performs the file I/O for the user. The server returns status
and data to the user.

In a typical DAP dialog, the first message exchange is of Configuration messages that
provide information about the operating and file systems, buffer size, and so on.
Attributes messages then supply information about the file. The Access Request
message typically follows to open a particular file. If data is to be transferred, a data

The Network Application Layer

6-3

stream is then set up. For both sequential and random access file transfer, one control
message sets up the data stream. After the completion of the file transfer, Access
Complete messages terminate the data stream.
.
|

Figuré 6-1 shows a DAP méssage e‘xc"'han:ge for sequential file retrieVéI._‘

Configuration Infor-

‘Configuration Message

mation (e.g. Buffer Size,
OS, File System, DECnet -

Remote NodeMessage
) Description

Messages
.

~©
S

User Node Message
_‘
Description

o
>

Phase No., and DAP

Version Number)

Configuration Message

Configuratibn

File Characteristics (e.g.

Type, Block Size and

Attributes Message

Record Size.
b

Information Returned

|
Access Message

Access Request

Attributes Message

Returned

Acknowledge Message _

Request Start of Data

Message

Acknowledge Message

Control (Get) Message

Transfer and declare

|
|

-~ Data Stream Established

mode of Transfer
v

Record 1

-«

o
o

Data Sentin Records

.
Request to Terminate file
access

Record N

-~
'

Status Message

Access Complete Message
|
Access Complete Response

)

End of File Detected
.

Request Completed

Successfully

Figure 6-1: DAP Message Exchange (Sequential File_Re_triéval)

6-4

The Network Application Layer

File Opened

Control (Initiate Data Stream)
Set up Data Stream

Actual File Characteristics

6.1.3 DECnet Remote File Access Facilities
- DECnet implementations currently access remote files using the folltowing facilities:
File Access Listener (FAL). FAL receives user I/O requests at the remote node
and acts on the user’s behalf. This is the remote DAP-speaking server process.

Network File Transfer (NFT). This interactive utility operates at the user

level. It interfaces to a DAP-speaking accessing process to provide DAP functions.

NFT provides network-wide file transfer and manipulation services..

‘Record Management Services (RMS). RMS is the standard file system for
many

DIGITAL’s

operating

systems.

particular,

DECnet-VAX

transparently supports RMS. RMS can transmit and receive DAP messages over
logical links.

If an RMS file access request includes a node name in the file

specification, RMS maps the access request into equivalent DAP messages. These

DAP messages are sent to a remote FAL to complete the request.

To the user,

remote file access is handled identically to local file access, except that a remote
node name and p0551b1y access control 1nformat10n is necessary for remote file
access.

Network File Access Routines (NFARs).

callable subroutines.

N FARs are a set of FORTRAN-

NFARs become a part of the user process;

-with FAL, using DAP, to access remote files for user applications.

they cooperate
RSX DECnet

\\__’/'/

uses N FARs to provide DAP functions.

VAX/VMS Command Language Interpreter VMS commands pertaining to
file access and manipulation interface with RMS to provide network-wide file

access; hence no separate NFT utilityis requiredin VAX/VMS.
Network Management modules.

Network Management modules use DAP

services to obtain remote files for down-line loading other remote nodes and to

transfer up-line dumps
for storage.

Figure 6—_2 shows node-to—node file transfer using DAP.

The Network Application Layer

6-5

' REMOTE NODE

LOCAL NODE

O@‘

UserlLayer

The accessing program:
-

NFT Utility
A VAX/VMS Command

A user-written
__program

‘Network Application Layer

Network Application Layer

DAP- speakmg server:

DAP-speaking routines:

File

NFARs, or Routines in

- the local file sy__stem<_f
(e.g. RMS)

Local

File
System

Access

Listener

- Session Control Layer

~ Session Control Layer

End CommumcatlonLayer

| End Communication
Layer

Routmg Layer

'Routing Layer

DataLink Layer

Data Link Layer

Physical Link Layer

Physical
Link Layer

(e.g. FCS or

( |

Router

- Router
.

Legend

NFT

- Network File Transfer UtilityTM

FCS

- File Control System
- Record Management Services

NFARs - Network File Access Routines
-DAP

- Data Access Protocol

Router - DECnet node performing Swutchmg

/l——-' " A Physical Communication Link
A path through the network

- DAP Message flow

Figure 6-2: File Transfer Across a Network

6-6

Remote

File

System

(e.g.FCS or
RMS)

RMS)

RMS

The Network Application Layer

6.2 Network Virtual Terminal Functional Description
The Network Virtual Terminal (NVT) service consists of the following:

A functidnal model of a terminal (virtual terminal) for access through the

A set of protocols for communication between a host node, with an application
“program running, and a server nede, with a terminal directly attached.

network. This model is known as the network command terminal.

The NVT is aSsociated with the Terminal Software Architecture, which also defines
nonnetwork aspects of terminal services.

- The Network Virtual Terminal service provides these functions and features:

-Distributes terminal-handling functions between two systems.
‘Supports heterogeneous host systems. Different operating systems can cooperate
via common protocols, and a host operating system can manage a terminal in its

own way, regardless of what operating system runs in the server.

Allows a server to connect to a specified host or a host to a specified remote
terminal.

Provides terminal input/output and characteristics management functions at the
operating system services level. The description of the command terminal

protocol includes a list of these functions (Section 6.2.2).

Offers standard terminal services, featuring good performance and device

independence. Optionally, offers methods of controlling the terminal’s behavior
in considerable detail.

Supports high availability implementations. The protocols contain functions that
allow the restart of interrupted communication.

The virtual terminal protocols are otganized into two sublayers of the Network
~

Application layer, as a provision for future enhancement. Figure 6-3 depicts the
organization of the Network Virtual Terminal service.

The Network Application Layer

6-7

SERVERNODE

HOSTNODE
UserLayer

Command Processor (e.g.
DCL)

User application

Server

Network Application Layer |

Terminal

Services

Command

Commang

Terminal Module [~ ~|TM Terminal Module

|
Terminal Communication Module [~ =TM unication Module

Session Control Layer

Host

Terminal

’ Services

Session Control Layer

End Communication Layer
Routing Layer

End Communication Layer
1

Routing Layer

Data Link Layer

Physical Link Layer

Network Application Layer

RN Terminal Comm——
|

oute r

Router
.

DECnet Network

Legend:

Router - DECnet node performing Switching
B

A Physical Communication Link

eiana

A path through the network

<-»

NVTProtocol Flow

Figure 6-3:

Network Virtual Terminal Service

Section 6.2.1 describes the Terminal Communication Protocol, which controls the

connections between applications and terminals.

Section 6.2.2 describes the

Command Terminal Protocol, which provides functions for terminals to communicate

with Command Language Processors and application programs..

describes the operation of the Network Virtual Terminal service.

Section 6.2.3
.

6.2’.1 The Terminal Communication Protocol
The base protocol for the Network Virtual Terminal

Communication Protocol.

Virtual Terminal service, known as the foundation layer.

6-8

The Network Application Layer

service is the Terminal

It is in the lower of the two sublayers of the Network

The Terminal Communication Protocol has responsibility for making and breaking
connections between applications and terminals. In effect, it extends DNA Session
Control layer services by establishing logical links between endpoints that are
specific to terminal services. The endpointin the host system is called a portal; the
other end,in the server system,is called a logical terminal.
A portal corresponds to a remote terminal identifier in the host, and a connection

binds that identifier to an actual terminal.
binding.

Thus, the NVT connection is called a
|

In the future, it may be desirable to add new models, as alternatives to the network
~command terminal. Such new models might require their own protocols. To provide

for this, the logical terminal, binding, and portal share a state called a mode. Through
mode management, the terminal communication protocol is able to select the
appropriate higher level Virtual Termmal protocol and model.
Table 6-2 shows the Terminal Communication Protocol messages types and their

functions.

TABLE 6-2:

Terminal Communication Protocol Messages

Message
-

Function

Bind Request

Requests a Binding;

identifies version and type of sending

system.

‘Rebind Request

Unbind

Requests a rebinding (reestablishes broken communications,

for high availability implementations).

_y |

Requests that a binding be released.

Bind Accept

Accepts a Bind request.

Enter Mode

Requests entry of a new mode (the only mode currently
defined is command mode).

This selects the command

terminal protocol as the higher level protocol.

Exit Mode

Confirm Mode
No Mode

Requests that the current mode be exited.

Confirms the entry of a new mode.
Indicates that the

requested

mode

not available

confirms an exit mode request.

Data

- Carries Data (i.e. command terminal protocol information).

6.2.2 The Command Terminal Protocol
The Command Terminal Protocol provides access to the functions of the network
command terminal model.

The network command terminal offers a set of functions oriented in-ainly toward
command line input and output, but general enough to support a broad range of video
and hardcopy terminal applications. These functions include:

Read a line of input, with echoing and operator editing performed by the server
system.

Line terminators may be individually specified. The host software has

considerable control over echoing and operator editing, 1nclud1ng the enabling
and disabling of each feature individually.

The Network Application Layer

6-9

“Accept input even if the program has not 1ssued a read request save.it until a
read request arrives (typeahead)

| Input or take action in response to, certain Charact'ers immediately as the keys
are

struck

(out-of-band character

individually spec1fied

These characters may be
-

processmg)

Write a string of output characters. Writing may proceed concurrently with
reading, subject to certain synchronization options.

The terminal operator can

cause output to be discarded.

'Recognize ANSI standard escape sequences on input and 'eutput.
Cancel a read request.

" Read and set termmal dev1ce characteristlcs

Determine how many characters are in the typeahead and input buffers“ |

combined.

Clear the typeahead and input buffers.

Table 6-3 shows the Command Terminal Protocol message types and their functions.
TABLE6-3:

Command Termlnal Protocol Messages

Message

Functlon

Initiate

Carries initialization information, as well as protocol and

implementation version numbers.

~ Start Read

| -Request_s that a READ be issued to the terminal.

- Read Data
|

Carries input data from te'rrninal'on completion of a read
|

Out-of-Band .'

Unread
Clear Input

request.

~ Carries out-of-band input data.

Cancels a prior read request.
| | Requests- that the input and type'ahead buffers be cleared.

Write o

"Requests the output of data to the terminal.

Write Complete

‘Carries write completion status.

Discard State

Carries a‘ change to the output discar‘dr state due to a
terminal operator request (v1a an entered output-discard

character).
Read Characteristics

Requests terminal characteristics.

Characteristics,,

Carries terminal characteristics.

Check Input
|

Requests input count (number of characters in the typeahead
and input buffers combined).

Input Count

Carries input count as requested with Check Input.

Input State

Indicates a change from zero to non-zero or vice-versa in the |

number of characters in the input and typeahead buffers
~combined.

The Network Application Layer

Y
.

6.2.‘3 NVT Operation
In a typical network command terminal session, a terminal management modulein a
server system with an active terminal requests a binding to some host system. It does

this by invoking a terminal communication services function.

(This typically takes

placein response to a command such as Set Host from the termmal operator.)

The terminal

communication services

module

initiates

a logical

link

counterpart in the host system, using DNA Session Control layer services.

its

discovering the incoming logical link, the host module allocates a portal, thus

beginning the formation of a binding. The host module then accepts the logical link.
Once the logical link has been formed, the server module sends a Bind Request
message to the host. A terminal management module in the host, perhaps acting on

behalf of the host’s login process, recognizes the binding request, and causes the
terminal communication services module to send a Bind Accept message.

Once the b‘indingohas been formed, the host takes the initiative and readies the
binding for the command terminal protocol (enters command mode) with a further

exchange of messages.

This takes place in connection with the first terminal I/O

request from the Login process.

The respective protocol modules can now begin

speaking the command terminal protocol, by each sending an Initiate message to the
other. After 1n1t1a11zatlon the dialog of remote terminal service requests and
responses ensues.
Terminal service requests normally originate with an application program in the host

system. The application program issues requests to host operating system terminal
services. Host terminal services issue corresponding requests to the host protocol

module. The server protocol module reproduces those requests remotely and reissues

them to the server termmal services.
Termmatlon of the session can come about in a number of ways.

A typical orderly

responds by releasing its resources. Finally, the host disconnects the logical link.

r_,-"
o

terminal communication services module sends an Unbind request.

shutdown begins with a Logout by the application in the host. As a result, the host
The server

Figure 6-4 shows NVT message exchange .

The Network Application Layer

6-11

Host Node Message

Server Node Message

- Messages

Description

)

" Accept Binding

Bind Accept Message

. Establish Command

_Enter Mode Message

- Mode

. specified Host

Bind Request Message

~ Request a binding with a

‘Server enters command

Confirm Mode Message

Initiate Message

Initialize command

terminal protocol

- mode

* Initialize command

terminal protocol

Begin terminal service
dialog
“Issue Read request

Start Read Message
Read Data Message

Host terminal software
notified of out-of-band

' Out-of-band Message

- Accept and process a line
of input

User types an out-ofband character

character

.‘Dialq_g continues

Dialog continues

- Application logs off.
‘Release host resources

Unbind Message

Release server resources.

~ Figure6-4: NVT Protocol Message Exchange

6-12

The Network Application Layer

6.3 X.25 Gateway Access Functional Description
X .25 Gateway Access provides the following functions and features:
®

Supports communication between user-written programs in a host DECnet
system and modules in a non-DECnet system over an X.25-based public data

network. (Two DECnet systems may communicate over an X.25 network using
standard DNA protocols without recourse to this gateway function. See Section

3.3.4.3 for a description of how this is accomplished.)

~ ®
®

Supports all facilities and modes of operation definedin CCITT Recommendation

X. 25 W1th the exception of Datagram mode and the Delivery Confirmation bit.

Supports user access to both Permanent Virtual Circuits (PVCs) and Switched
Virtual Circuits (SVCs). Switched Virtual Circuits are also referred to as virtual
|

calls.

Permits user-written programs to access PVCs or issue virtual calls from any
DECnet system contammg an X.25 Gateway Access module. The user program
does not have to reside in the DECnet system thatis the DTE (Data Terminal
Equipment) directly connected to the X.25 network.

Permits incoming virtual calls to be directed to a user process residing at any node
in the DECnet network. A Network Management function maps call data from
incoming virtual calls to a DECnet process description so that the call may be
directed to the correct process.

~®

Permits user access to all X.25 packet fields such as more data, qualified data, ete.

Recovers from transient errors, and reports fatal errors to the user.

X.25 Gateway Access is designed to make the full capabilities of the CCITT X.25
Packet Level interface available to user programs residing anywhere in a DECnet
network. To accomplish this, X.25 Gateway Access communicates with the DECnet
system that is the DTE on the X.25 network over a DECnet logical link (the DTE
system is often called the Gateway system). X.25 Gateway Access protocol messages
are exchanged over the logical link to make the facilities of the Packet Level interface
available remotely.

" Section 6.3.1 describes the messages exchanged by the X.25 Gateway Access Protocol,
Section 6.3.2 describes the operation of X.25 Gateway Access, and Section 6.3.3
describes the modules that cooperate to provide X.25 Gateway Access.

6.3.1 X.ZS Gateway Access Messages
X.25 Gateway Access uses the messages listed in Table 6 4 to accomplish remote
access to the X.25 network.

The Network Application Layer

6-13

TABLE 6-4:

X.25 Gateway Access Messages

Message

Function

Open

Requests the use of a permanent virtual circuit.

Open Accept

Accepts a permanent virtual circuit Open request and
allocates the PVC to the Gateway Access user.

Open Reject

Rejects a Permanent Virtual Circuit Open request, refusing

allocation of the PVC to the Gateway Access user.

Outgoing Call

Requests that an outgoing Virtual Call be placed byissuing
a Call Request Packet to the X.25 network

Call Reject

Indicates rejection of an outgomg or 1ncom1ng virtual call for

lack of resources.

Incoming Accept

~ Indicates acceptance of an outgoing virtual call.

Incoming Call

Indicates an incoming virtual call has been received.

Outgoing Accept

Accepts a p.reviousl'y received ineoming virtual call.

Clear Request

Requests that a virtual call be Cleared.

Clear Indication

Indicates a clear request packet has been received from the

X.25 network.

Clear Confirm

Indicates a clear confirmation packet has been received from

the X.25 network.

Requests that a virtual circuit be reset.

Reset Request
Reset Indication

Indicates that a reset indication packet has been received

from the X.25 network. |

Reset Confirmation

~ Indicates.a reset confirmation by elther the user or the X.25

network.

Reset Marker

Marks the place in the stream of Data messages Where a
reset occurred.

Data

Contains Data pack,ets'o-ut.bound fro:r’n or inbound to a user.

- Interruptv

Contains Interrupt packet data outbound from or inbound to
a user.

Interrupt Confirmation

Confirms the receipt of an Interrupt message (used for flow
control).

Indicates that a PVC entered a no-communication state.

No Com
No Com Seen

Indicates that the user of a PVC acknowledges a No Com

message, and thus has been notified that there have been one
or more failures of the X.25 network (e.g. Restarts).

6.3.2 X.25 Gateway Access Operation
Three processes are involved in X.25 Gateway operation: the user process, the X.25

Gateway Server process, and the foreign X.25 DTE process. The user process accesses
the X.25 Packet level functions by issuing calls on the X.25 Gateway Access module.
These

calls

are

translated

into

X.25

Gateway Access

protocol

messages

and

transmitted over a DECnet logical link to the X.25 Gateway Server process. The X.25
Gateway Server transmits and receives X.25 packets to and from the X.25 network.
These packets in turn flow over an X.25 virtual circuit to the foreign X.25 DTE
process.

6-14

The Network Application Layer

In a typical X.25 Gateway Access Protocol dialogue, the first message exchange is
either to open a Permanent Virtual Circuit via an Open message or to place a virtual
call using an Outgoing Call message. In the case of a virtual call, when the foreign
DTE process has accepted the call byissuing a Call Accept packet, t_he user process is

informed by the receipt of an Incoming Accept message. The user process may then
exchange data with the foreign DTE process by sending and receiving Data and
Interrupt messages. When the user wishes to terminate the virtual call, a Clear
Request message is sent to the X.25 Gateway Server, which clears the call byissuing
a Clear request packet to the X.25 network. The user process is then 1nformed of the
termination of the call by receipt of a Clear Confirm message.

“Flgure 65 shows X.25 Gateway Access protocol exchange for Switched Virtual
Circuit operation, where the virtual call is initiated by the user process on the
DECnet system.

User Node Message
Description

Messages
-

| Outgoing Call Message

Request to place a Vnrtual

CallTM

‘Gateway Node Message
~ Description

Gateway Server process

issues Call Request packet

to X.25 network

Call Accept Packet

In;ommg Accept Message

received from X.25

network

Request to send a Data

Packet

Data Message

Data Packet sent on

virtual circuit

e
e

Data Message

-«

Userreceives Datasentby
foreign DTE process

User receives Interrupt

. Data

“

Data packet rec_eived'

Data Message

from virtual circuit

Interrqpt Message

~ Interrupt Packet received
from virtual circuit

“Interrupt Confirmation Message

~ Request to terminate

Clear Request Message

(clear) Virtual Call

Virtual Call is terminated

<Clear Confirm Message

Clear Packet issued to
X.25 network

Clear Confirmation

Packet received from

X.25 network

Figure 6-5: X.25 Gateway Access Message Exchange

The Network Application Layer

6-15

6. 3 3 X. 25 Gateway Access Modules

DECnet 1mplementat10ns use the followmg fac111t1es to prov1de the X.25 Gateway'

Access service:

X.25 Gateway Access Module This module receives user X.25 requests and

communicates ‘with the DECnet system thatis the DTE dlrectly connected to the
X 25 pubhc data Network (the Gateway system)

X.25 Gateway Server Module. This module residesin the DECnet system that

is the DTE on the X.25 network (the Gateway system).
This module
‘communicates with the X.25 Gateway Access module over a DECnet logical link
and acts on requests made by the X.25 Gateway Access module and the X.25
network. It uses the facilities of the m.25 Packet Level Module to gain access to
the X.25 network.

X.25 Packet Level Module. This module resides in the Data Link layer of the

Gateway system and acts as the DTE on the X.25 network. It uses the X.25 Frame
level module for its connectwn to the X.25 DCE ‘This module is described in
Sectlon 2.3.2.

X 25 Frame Lével Module. This module also residesin the Data Link layer of
“the Gateway system. It provides the X.25 Frame level protocol in order to
communicate with the X.25 network’s DCE. This module is described in Section
2.3.1.

Figure 6-6 deplcts the relat10nsh1p of these modules to the X.25 network and to the
foreign DTE:

6-16

'The Network Application Layer

The accessing program:

program |

Network Applicat‘on Layer

~ Network Application Layer

X.25 Gateway Access

X.25 Gateway Server Module @

’

The cooperating program

Foreign DTE Process

~ Auser-written

User Layer

FOREIGN NODE (DTE)

GATEWAY NODE (DTE)

~ LOCAL NODE

‘Module

Session Control Layer

End Com'mUnicatibn‘;Layer
Routing Layer \'; ]

inkLaver

End Confinunication Layer | :
:
__Routing Layer

Network La;'yer
| X.25"Packet"Level

Data Link Layer

Session'Control Layer

Physical Link Layer \

ApyDNA

X.25 PackgtLevel

Protocol |

| X.25Frame Level

— i

Datalink

—

;

Physicallink Layer ]

X.25 Franfe Level

1|1

PhysicalLink Layer

X.25 Interfac

}R.outer

Router |

DCE

DCE .
.

DECnet Network

X.25 Packet Switching Networ

. Legend:

DTE

- Data Terminal Equipment

DCE

- Data Circuit Equipment

Router

- A DNA node performing Switching

- Incoming Call mapping Database

= ~ APhysical Communication Link
..... ...

Apath through a network

—.—»

Data Flow through X.25 Gateway Access

| Figure 6-6: X.25 Gateway Access Operation |

| The Network Applicatioh Layer

6-17

6.4 SNA (Systems Network Architecture) Gateway Access Functlonal
Descrlptlon |
SNA Gateway Access provides the following functions and features:

Supports communication between user-written (and a’ number of DIGITALwritten) programsin a hostDECnet system and modulesin an IBM host system
:
over an SNA network

Supports programs in DECnet systems whlch act as SNA Secondary Log1ca1 Units

- (SLUs) of any of the defined SNA Log1ca1 Unit Types (LUs)

- ®

Allows user programs full access to the facilities prov1ded by the SNA
Transmission Control and Data Flow Control layers.

° Communlcates W1th the SNA network as a Physical Unit Type 2 (PU2)
e Permits user—wr_lttenprograms to-~part_1c1pate in SNA sessions from any DECnet
system containing an SNA Gateway Access module. The user program does not

“have to residein the DECnet system thatis actmg as the PU2 directly connected

to the SNA network.

[ | Manages the SNA session pacing automatlcally for the user program
®

Recovers from transient errors and reports fatal errors to the user

SNA Gateway Access is designed to make the full capablhtles of the SNA Data Flow

Control, Transm1s51on Control, and Path Control layers available to user programs
residing anywherein a DECnet network. -To accomplish this, SNA Gateway Access
-communicates with the DECnet system thatis the PU2 on the SNA network over a

- DECnet logical link. SNA Gateway Access Protocol messages are exchanged over the
loglcal 11nk to make the fac111tles of SNA sessions avallable remotely.

Sectmn 6.4.1 descrlbes the messages exchanged by the SN A Gateway Access protocol
Section 6.4.2 describes the operation of SNA Gateway Access, and Section 6.4.3
describes the modules that cooperate to provide SNA Gateway Access. |

- 6.4.1

SNA Gateway Access Protocol Messages

SNA Gateway Access uses the messages hsted in Table 6-5 to accomphsh remote
access to the SNA network.
-

6-18

The Network Application Layer

TABLE 6-5:

SNA Gateway Access Messages

Message

Connect
-

Function

- Listen

Bind Data

Requests that a Session be established with a Primary

Logical Unit (PLU) inan SNA host.

Waits for a Session to be established by a PLU.
Indicates that a BIND has been received for a session

solicited w1th Connect or Listen.

Bind Accept

Requests that the session being established with the BIND
|

be accepted and placed in the running state.

~ Normal Data

Carries data on the SNA session’s normal flow to and from
the PLU.

Interrupt Data

Carries data on the SNA session’s expedited flow to and from
|

the PLU.

/,/’

Disconnect
|

Requests orderly session termination.

Disconnect Complete
|
|

Abort

Indicates normal
successfully.

session

termination

has

completed

Requests abnormal termination of a session, or indicates an
UNBIND has been received from the PLU.

'6.4.2 SNA Gateway Access Operation
Three processes are involved in SNA Gateway Access operation: the user process, the

SNA Gateway Server process, and the IBM host application process. (Note: IBM hosts
are called PU5 nodes in SNA terminology.) The user process accesses the SNA SLU
functions by issuing calls on the SNA Gateway Access Routines. These calls are
translated into SNA Gateway Access protocol messages and transmitted over a
DECnet logical link to the SNA Gateway Server process. The SNA Gateway Server

transmits and receives SNA Basic Information Units (BIUs) by accessing the

functions of an SN A protocol emulator, which connects directly to the SNA network as
a PU2 and contains the functions of the SNA Path Control and Data Link Control
(SDLC) layers.

In a typical SNA Gateway Access protocol dialog, the first message exchange is to

establish a session with a PLU in the IBM host, by exchanging Connect and Bind
Data messages. The session is completely established when the Bind Accept message
" is successfully sent.

The user process may then exchange data with the PLU by

sending and receiving Normal Data and Interrupt Data messages.

When the user

- wishes to terminate the session, it informs the PLU by using the appropriate Data

Flow Control Request Unit (RU).

The PLU may then terminate the session by

sending an UNBIND, which results in session termination with an appropriate

- reason code to the user process.

Figure 6-7 shows SNA Gateway Access Protocol exchange for a session requested by
the user process (the SLU):

‘'The Network Application Layer

6-19

‘proposed by BIND

User Receives Normal
flow Data

‘

.- Gateway Node Message
- Description
|

Gateway System issues an
. INIT-SELF message to the
-

SNA network

: .Bind Data Message

BIND received from PLU

Bind Accept Message
—

_ User accepts session as
|

S Connect Message
'

Request to establish an
. LU-LU session
.

Messages

User Node Message
Description

toPLU

Normal Data Message

Gateway accepts session
.

;

| Normal Flow RU received
|

from PLU

Normal Data Message

User Sends normal flow
RU to PLU

‘

Interrupt Data Message

Expedited flow RU
received from PLU

User sends expedited

Interrupt Data Message

flow Data to PLU

- User senses session termination

o
h

~ AbortMessage
~PLUsends UNBIND to
terminate session
:

'Figure 6-7: SNA Gateway Access Mess_age Exchahge
6.4.3 SNA Gateway Access Modules
DECnet 1mplementat10ns use the followmg facilities to provide the SNA Gateway
- Accessservice:

“e
~

SNA'Gateway Access Mddule. This module receives user SNA session requests

and communicates with the DECnet system that is the Physical Unit (PU)
directly connected to the SNA network (the Gateway system)

‘@ SNA Gateway Server Module. ThlS module resides in the DECnet system that
is the PU on the SNA network (the Gateway system). This module communicates
with the SNA Gateway Access module over a DECnet logical link and acts on

requests made by the SNA Gateway Access module and the PLU.

It uses the

facilities of an SNA protocol emulator to gain access to the SNA network.

6-20

SNA protocol emulation. This set of modules resides
in the Gateway system
and performs the functions of SNA Path Control and Data Link Control. It
connects to the SNA network as a Physical Unit Type 2 node. The SNA protocol
emulation modules also perform the functions necessary to communicate with the
SNA network’s System Service Control Point (SSCP).

The Network Application Layer

Figure 6-8 depicts the relationship of these modules to the SNA network and to the
IBM host application (the PLU):

LOCAL NODE

GATEWAY NODE

IBM HOST NODE

UserLayer

PLU

The accessing program

(SLU):
SNA Remote lob Entry

The Cooperating program:
(e.g. CICS, IMS, JES2,

3270 Emulation

JES3, etc.)

A user-writtent

program

Y
Network Application Layer

P
]

SNA Gateway Server Module
4
|
\

Session Control Layer

Sessidn Control Layer

End Communication Layer
‘

Data Link Layer

|
Network Application Layer

SNA Gateway Accss
Module
!

Routing Layer

)

Data Flow Cont[rol Layer

Path

End Cgmmunication Layer C?) ntrol

/Routing Layer

Path Control Layer

;

+ Datalink Layer

]

Data L\ink

Data Link Layer

/ Physical Link Layer | €Ontrofy

Physical Ljnk Layer

Physical Link Layer Y

Transmission Ctl. Layer

(PU2}.

[
NAU Servicesi.ayer

Router

PU4
PU4

DECnet Network

.
SNA Network

Legend:

SLU

- Secondary Logical Unit

PLU

- Primary Logical Unit

PU2

- Physical Unit Type 2

PU4

- Physical UnitType

Router

- A DNA node performing Switching

|
4

A Physical Communication Link

«oooeeee.

A path through a network

—.—»

Data Flow through SNA Gateway Access

Figure 6-8: SNA Gateway Access Operation

The Network Application Layer

6-21

\\\ e-

Network Application

Session Control

End Communication
Routing

Data Link
Physical Link

Chapter 7

\._/J

)

Network Management

Network Management allows system managers to control and monitor network

operation.” Network Management also provides information for use in planning the
evolution of a network and correcting network problems.
The Network Management design has three outstanding characteristics:

° Both programs and terminals can access and control a DECnet network via a set
of functionally discrete calls and commands.

-® Control over a DEChnet network can be either distributed or central. Distribution
|

of control can be either partial or complete.

Network Management is a set of primitive functions or “tools.” The system or
network manager can fashion them into a management system that meets his or
her specific needs. This allows the managers of a network to construct their own

network management philosophy.

Most of the Network Management modules reside in the Network Management layer.

In addition to these, however, there are Network Management modules in the User

and Network Application layers. Also, one Network Management module, the Event

Logger, has a queue residingin each lower layer. This chapter dlscusses all Network
» Management modules regardless of where they reside.

~ Since Network Management is modular a DECnet system is not requlred to

1mplement the architecturein its entirety.

7.1

Network Management Functional Description

Network Management performs the following functions:
®

Loading and dumping of remote systems. For example, a system manager can
down-line load an operating system into an unattended remote system.

Network Management

7-1

Changmg and examining network parameters. For example, an operator can

' change circuit costs or node names.

L) ."Exammg network counters and events that indicate how the networkis
performing. For example, the Event Logger automatmally records s1gn1ficant
network events.

Testlng links at both the data link and logical link levels. For example a

- system manager can send a test message that loops back to its origin from a
'specific pointin the hardware connection.

an
Setting and displaying the states of lines and nodes. For example,
operator can reconfigure a network by turning nodes and links on or off.

7.2 Network Management Operation
This section summarizes the functions of each Network Management component and

- describes the operation of the major Network Management functions.

~7.2.1 Components
The Network Management components are as follows:

User Layer

® Network Control Program (NCP). NCP is a utility at the user level that

interfaces with lower level modules. NCP has.a standard set of interactive
terminal commands that each DECnet implementation uses.

Network Management Layer
®

Network Management Access Routines. ' These routines provide generic
- Network Management functions. The routines communicate across logical links

with the Network Management Listener using: the Network Information and

Control Exchange (NICE) protocol.

'®

Network Management Listener. The Network Management Listener receives
Network Management comimands from the Network Management level of remote
' nodes via the NICE protocol. In some 1mplementat10ns it also receives commands
from the Network Management Access Routines via the NICE protocol. The
Network Management Listener passes function requésts to the Local Network
Management Functions.

Local Network Management Functions. This component takes function

requests from the Access Routmes translatmg the requests1nto system- -

dependent calls.
®

Link Watcher. Sensing service requests on a link from an adJacent node the

Link Watcher haridles state changmg for automatic remote load, dump, and

trigger functlons

7-2

Network Management

Maintenance Functions. The Maintenance Functions provide the actual protocol operation to support system maintenance functionssuch as down-line
load and data link loop testing.

® Link Service Functions. The higher level Network Manag_ement modules

interface to the Link Service Functions for services that require a direct data link
(bypassing the Session Control, End Communication, and Routing layers).
® Event Logger. The Event Logger records significant events occurring in the
lower layers. DNA specifies event types in the Network Management interface to
each layer. An event processor within the Event Logger takes raw events queued
" in each layer and records events of the types specified by the system and system
manager. Using the Event Logger protocol, an event transmitter can inform
of event occurences. Events travel to a specified
at other nodes
event receivers
'
e file or monitor output.
for console,
* sink nod

\\w/,

in the
Figure 7-1 shows the Relationship among Network Management| components
:
|
User and Network Management layers.

Network Application Layer

' ®

Loopback Access Routines and Loopback Mirror. For logical link loopback
tests, the Local Network Management Functions can interface to the Loopback
Access Routines, which use the Loopback Mirror protocol to loop test messages
a remote Loopback Mirror.
from

Network Management

- 7-3

/‘
\\-«9”//
.

User

_NCP

Program
|

Link

Watcher

UserLayer

Network
Management

commands

Protocol

Network

Protocol

<% -----1

Management

p----- —»1

Network

Management

Access Routines

Network

nodes

Management |[€-------

Listener

commands
to other
‘nodes
\4

Local Network Management Functions

Maintenance

Functions

Events to other »

Events from

nodes

Link Service

Functions

other nodes

Event Logger

€

Network

Management
Layer

Data Link layer

communication
-

Control over lower

level functions

(examine line state,
turn on routing,
etc.)

Lower

Layers

System dependent calls
to application layer and

local operating system
functions (file access,

logical link loopback,

Read event
queues

timer setting, etc.)

KEY:

Figure 7-1:

7-4

——=—=-3

Peer protocol interface

===

Network managementinterface

l Client interface

Relatlonshlp among Network Management Components
at a Single Node

Network Management

'7.2.2 Remote Loading, Dumping, Controlling, and Link Loopback
Testing Functions

Network Management uses the Maintenance Operation Protocol (MOP) to perform
remote loading, dumping, controlling and testing. (Refer to Section 7.4). The target
(the node being loaded, dumped, eéc.) or the executor (the adjacent node executing the
requests)or a remote command node can 1n1t1ate the function. Figure 7-2 illustrates

the down-line load request operation.

Link loopback testing is a procedure for isolating faults in a physical connection
between two nodes. Link loopback tests consist of sending out test messages that are
looped back at some point. By changing the loopback point, an operator can precisely
locate problems.

There are three methods of performing link loopback tests:
1. Using NCP commands, the operator can set certain line devices to loopback mode,

and then perform the loopback tests. This “automatic” method can be useful in
testing unattended remote nodes.

To loop a message from another hardware point, such as a modem, the operator
must set up a loopback device manually. This can be done by throwing a loopback
switch on a modem or connecting a loopback plug in place of a modem. The
|
operator can then execute the loopback test using NCP commands.

3. By not setting a hardware loopback device and using link loopback NCP
commands, an operator can cause loopback test messages to loop back from the
Mamtenance Functionsin the Network Management layer of the adJ acent node.

Figure 7-3 illustrates link loopback tests.

Network Management

7-5

1. Target-initiated Request
Executor Node
Target Node

Link

Watcher

Loéd
Requester

Local
Network
Management

2. Opératoreinitiated Request from a Remote Command Node
Command Node

‘Network
Management
Access

Routines

Network

Logical Link

Executor Node

Network

Control

| Managemen

Program

Listener

Local
Network
1

Target Node

Managemen

Functions

Console
Server

Figure 7-2: Down-Line Load Request Operation

7-6

Network Management

Functions

A. Direct Line Loopback, hardware looped

Exécutor Node
User
Layer

N.M.

Network

Management

w P

Maintenance Functions

Data

Link

Physical
Link

) v¥

Local N.M. Functions

N-M.

Listener

Layer

lep|

Access Routines

']
1

Hardware Loopback

Device

B. Direct Line Loopback, software looped
Executor Node

User

Layer

Target Node

User

NCP |

Layer

[

N.M.

Network

Management
Layer
l

Access Routines

Listener

Networ

)

v+
: v?
- Local N.M. Functions

Management
‘| Layer

Maintenance Functions

Data

Link

Physical

Link

}

: [

Maintenance Functions

—

Data
Link

Physical

;

Link

.
|

Communication Hardware

Legend:

» control flow

~ - data flow

N.M. = Network Management

Figure 7-3: Line Loopback Tests

Network Management

7-7

7.2.3 Node Loopback Testing
Node loopback testing consists of sending test messages over a logical link to be looped
back at a specific point. The operator may set a hardware loopback device on a line,
line device, or modem between the executor and adjacent target. Alternatively,
software components can loop back test data. Different software components can be
used. For example, FAL, a user program, or the Loopback Mirror can loop data back
to their associated access routines. Figures 7-4 and 7-5 describe some types of node
loopback tests.

One type of node loopback test uses the Network Management Loopback Access
Routines and Loopback Mirror residing in the Network Application layer (Figure 74,
B and Figure 7-5, C). These modules communicate over logical links using the
Loopback Mirror Protocol.

Other loop tests use logical links, but do not use Network Management software. For
example, Figure 7-5, B shows a test message travelling from one user task to-another.

Figure 7-5, D shows a file transfer used as a logical link test.

By using the NCP SET NODE LINE command, an operator can set up a special loop
node path (Figure 7—4). In this case, if no hardware loopback device is set, the
adjacent node’s Routing layer loops back the test.

7-8

Network Management

A. Local to Loop Node Test, Sin?(Ie Node, using filesastest §
data with hardware loopback.

User

Modules

I Use;TaskJ

Net. Man.

—

Network
gt

N
File Acc. Rou.

‘

Session Cont. !\‘
End Comm.

hardware loopback.

[

Modules

] |[

s
Data Link
.
.

{

PhysicalLlink

]

T
\

Session Cont. :

tl_ard.
oop

Dev.

L _________

—

L. TIZZzZzzzzzZs pmian

[

-—1A;] lEop. Mirror
: ]

—t7

'
Func.
’ Loc. Net. Man.
_

Application l qup. Acc. Rou.

Hardware | '

Network

N L

7
R
~7

Comm.

T
!

Net. Man. Acc. Rou.

Management
g

—
1
!

Routing
i

: Acc. List.
:
File

messages with

User

Network

Apphcatuc?n I !
’

B. Node Test,' Single Node, using Ioogback mirror and test

L__-CZZICZZZZZZZZ-ZZ-ZIi-o

‘F
:

\
.
\

End Comm.
Routin

Physical Link

]

Comm.|

Hardware

/oy
o

Data Link

‘L L o

- Hard.
-

L0oOp

Dev.

L.ZZZzZZZZzZzZ3==—~

————————————————

,_3 ]
[pa—
g

C. Local to Loop Node Test, Two Nodes, using user tasks.
User
Modules

[ User Task ]

I User Task» ]

Net. Man.
Net. App.

Net. Man.

‘\‘

‘/

Net. App.
Session Cont.

Session Cont. |

End Comm.

‘/' ‘

Routing

'\‘

Data Link

Ph‘ysical Link

"‘L

Comm.

Hardware

User
Modules

End Comm.,

/!
,.1’

Routing

Data Link

Physical Link

‘l

ee-

R A\\

D. Local to Loop Node Test; Two Nodes, using loopback mirror and test messages.
|

User .

Modules
Network

niep

Net. Man. Acc. Rou.

Management

Modules
Network

| |oc. Net. Man. Func.

Network

User

‘Management
,

Application U?Op' Acc. Rou. [LooP. M'"_OTI

Network

Application

Session Cont. ‘L

,’T,

Session Cont.

End Comm.

)

L/, L/'

End Comm.

Routing

CEN

Data Link

)

Data Link

Physicallink
-

Comm.|

Hardware

“

/’ K

—

‘L

Legend

e e

—3Pp

Physical Link ~ ;

L b=

Le e e e e

control flow

—~-—p

_1_‘1

data flow

Figure 7-4: Examples of Node Level Testing Using a Loopba_ck Node

Network Management

- 79

B. Normal Local to Local, using user tasks.

A. Normal Local to Local, using loopback mirror.
.

Network
Management

User
Modules

Net. Man.
Net. App.

]

4
]
L

]

¥-

Physical Link

Data Link

User Task

]

Routing
Data Link

\
\

End

V)
Y

Routing

‘

Session Cont.
Comm. |

'
P
‘
-'
.
:
|
Networ
Application fioop. Acc. Rou.] .[Loop. errorJ_

Session Cont.
End Comm.

er Task

Net. Man. Acc. Rou.
:
Loc. Net. Man. Func.

rUs r

Comm.

. Hardware

Physical Link
Comm.

- Hardware

C. Normal Local to Remote, using loopback mirror.
User

User

Modules

NCP

Modules

Network

Net. Man. Acc. Rou.

Network

Manageme'nt

Loc. Net. Man. Func.

Mavnagem‘ent

Network
Application

Network |—L:)<‘>p. —=
Acc. Rou.J
Application
\

End Comm.
Routing

Physical Link

Comm.

‘Hardware

End Comm.
Routfng

Data Link

Session Cont.

Physical Link

Y l

ee e e e

D. Norm,al Local to Remote, using files as test data.

User
Modules

:
|
]Mirror
[ Lo;)p.

L L

e et b

/,’

—————————— —

//j

4/ ]

,’7

,fi |

' Task ]
I User

User
Modules

Wser TaskJ

Net. Man.

Network
Application
Session Cont.

End Comm.

Comm. |

Hardware

Data Link

\
1

Physical Link

]

Physical Link

l,'

Routing

Routing
Data Link

USSP

Legend:

—_)

control flow

‘/

‘

l File- Acc. L'SLJ
1*

Network
Application
Session Cont.

rF"‘e- Acf' ROU'J
\

;

data flow

Figure 7-5: Examples of Node Level Logical Link Loopback Tests

7-10

Network Management

\‘W/g

7.2.4 Parameters, Counters, and E‘vents
DNA specifies line, circuit, node, module, and logging parameters that Network

Management can access.

These are internal network values that are important

enought to the system manager to examine, and, in some cases, change. These values
are classified as characteristics or status. Characteristics generally remain constant

until changed
by a system manager. Status parameters reflect the condition of lines,
circuits, nodes, modules, or loggers.

For example, line state is a status parameter.

The action of the network can change line, circuit, and node states. The logging state
“is really a system-manager-controlled on/off switch.

Each parameter has a unique type number. Some examples of parameters are:
®

Circuit cost

o Nbde address
'®

Linestate

Node hops

Logging sink node

DN A also specifies line, circuit, node and module counters and events. These are
internal network variables that keep track of errors and network activity at each

)'//

layer.

The Network Management design specifies only counters and events that are useful
to the system manager. For example, certain Routing counters enable a system
manager to determine if there are too many packets being discarded by the network

. due to congestion. In this case, the manager can adjust parameters that control how
packets are routed in order to better balance traffic loads. Additionally, the manager
can decide whether or not he needs to order more communication facilities.

\\\

’ p)

The Network Management design limits the overhead to the network involved in
gathering statistics and errors as much as possible. For example, there are no
" Routing counters that detect bugs in the Routing code: Network Management
assumes the code is correct. The Routing layer algorithms themselves have been
designed to detect and recover from most failures which would disrupt the network
operation.

The Local Network Management Functions interface to network counters (see Figure
7-1). The Event Logger receives event information (Figure 7-1). Changes in counter
values trigger many events. The Local Network Management Functions return
counters in response to user level requests. On the other hand, the Event Logger, once
set up, records events for user use automatically.

Each DNA functional specification contains, if applicable, its Network Management

interface, including counters and events. The DNA Network Management Functional
Specification contains tables of all Network Management parameters, counters, and

events.

Network Management

7-11

7.3 Network Management Messages
The three Network Management protocols are:
1.

Network Information and Control Exchange (NICE) Protocol. This handles
most Network Management functions.

Event Logger Protocol. This handles event logging from remote nodes.

Loopback Mirror Protocol. This handles the node loopback function.

Table’ 7-1 describes the NICE messages.
Table 7-1:

NICE Messages

Description

Message
Request Down-line Load

Requests a specified executor node to down-line load a target node.

Request Up-line Dump

Requests a specified executor node to dump the memory of a target node.
Requests a specified executor node to trigger the bootstrap loader of a

Trigger Bootstrap

target node.

Test

Requests a specified executor node to perform a node, circuit, or line

Change Parameters

Requests a specified executor node to set or clear one or more Network

loopback test.

Managementparameters.

Requests a specified executor node to read a spemfied group of

Read Information

parameters or counters.

Requests a specified executor node to either read and zero or zero a

Zero Countérs

specified group of counters.

Requests a system-specific

System Specific

Response

Network Management function.

Provides request status and requested information in response to a

NICE request.

There is only one event logger message, the Event message. It provides information
when an event occurs.. This information includes:

The event sink (in other words, where the event is to be logged -- one or more of
console, file, or monitor)
The event type and class

The date and time the event occurred at the source node
The name and address of the source node

Whether the event related to a module, node, circuit, or line
Specific data concerning the event.

Table 7-2 describes the Loopback Mirror Protocol messages.
Table 7-2:

Loopback Mirror Messages

Message

Command

Response

7-12

Network Management

Description

Requests a loop test and sends the data to be looped.

Returns status information and the looped back data.

7.4 Maintenance Operation Functional Description
Maintenance operations are special, primitive functions that must be available
without the services of any DNA layers between Network Management and Data

Link. Thisis because one of the nodes involvedin the operation may be in a state

where it cannot support more than a minimal Data Link. This could be, for example,
a node that is being down-line loaded or up-line dumped.”

At least some maintenance operation is supported in all of the DNA compatible Data
Link protocols. DDCMP supports all functions in its maintenance mode. The X.25
frame level supports line loopbackin a special loopback mode. Ethernet supports all
functions as maintenance protocol types.

In most cases, maintenance operations are accomplished using the Maintenance

Operation Protocol (MOP). In the case of the Ethernet, loopback testing is done with a
standard protocol being incorporated into the Ethernet Functional Specification.

Maintenance Operation includes the following functions:
®

Down-line loading the memory of a computer system.

Up-line dumping memory contents, usually upon a system failure.

Loopback testing of the data link and/or its hardware components.

System console control for a remote and possibly unattended computer system.

7.4.1 MOP Messages

Table 7-3, following, describes the MOP messages.
Table 7-3: MOP Messages
|

Message

Description

without

Causes the contents of the image data to be loaded into memory at the
load address, and the system to be started at the transfer address.
Causes the contents of the image data to be loaded into memory at the

Request Memory Dump

Requests a dump of a portion of memory to be returnedin a Memory

Request Program

Requests a program to be sent.

Memory Load
- Transfer Address
Memory

Load

with

" Transfer Address

Request Memory Load

Dump Data message.

Request Dump Service
Memory Dump Data
Parameter

Load

load address.

with

Requests the next segment of image data in a loading sequence and

provides error status on the previous segment.

Requests a dump to be taken.
Returns the reqflested memory image in response to a Request Memory
Dump message.

Loads system parameters and transfers control tothe loaded program.

Transfer Address

Dump Complete

Signals completion of a requested dump.

Assistance Volunteer

Indicates willingness to perform dump or load service in response to a
Request Program or Request Dump Service message.

(continued on next page)

Network Management

7413

Table 7-3 (Cont.):

MOP Messages

Méssage

LoopData

Description
‘Contains data to be sent back in a loopback test.

- Looped Data

Returns the data from a Loop Data message.

Boot

|. Causes a system to reload itself if the verification code matches and the
system is willing to do so. May result in a down-line load.

Request ID

System"ID

Causes a system to send a System ID message.

Identifies the sending system by, for example, maintenance vérsion,

maintenance functions supported, processor type, etc.

Request Counters

Causes a system to send a Counters message.

Counters

Returns Data Link counter values in response to a Request Counters
message.

Reserve Console

~ Reserves a system’s console for dialog with the requester.

Release Console

Releases a console reserved with the Reserve Console message.

Console Command and
Poll
A

Sends command data to a reserved console and/or polls for response
data.

Console Responseand
Acknowledge

Command and Poll message.

Returns response data and/or acknowledges receipt of a Console

7.4.2 MOP Operation
| Acéording to functional requests from higher level modules, or in responée to MOP
messages received from remote systems, MOP sends appropriate messages within the

Data

Link

envelope.

MOP

messages

and functions

handle

all

message

acknowledgment, time-out, and retransmission functions.

The node being serviced (being down-line loaded, up-.line' dumped, controlled, or

tested) is called the target node. The node providing the services is called the executor

node. MOP messages pass alternately between the executor and target.

Ideally, a target system would have all programs necessary to process MOP messages
available in local main memory, Read Only Memory (ROM), or mass storage.
Practically, this is not always the case and programs must be down-line loaded. This
requirement to down-line load the desired function even extends to down-line load

itself, which must usually be loaded through stages of progressively more capable

| loaders.
In all MOP message interchanges it is the responsibility of the sender of the request
message to time-out and retransmit if a response is not received.

7.4.2.1 Down-line Loading
Either the target or the executor system can initiate a down-line load. In either case,
the target sends a Program Request message to tell the executor what is needed and to
indicate its willingness to cooperate.
The major message exchange consists of
Memory Load messages from the executor and Request Memory Load responses from

the target.

The sequence is completed when the host sends a Memory Load with

Transfer Address or a Parameter Load with Transfer Address message.
On an Ethernet data link, a target can multicast its program request and select an

- executor from those responding with an Assistance Volunteer message.

7-14

Network Management

7.4.2.2 Up-line Dumping

Up-line dumping is very similar to down-line loading. The target uses the Request
Dump Service and Memory Dump Data messages. The executor controls the process
with the Request Memory Dump and Dump Complete messages.

7.4.2.3 Link Testing
MOP tests the data link by looping a test message from a point in the physical
connection.

By moving the loopback point and isolating components, the user can

diagnose problems. The active side (the executor module controlling the test) sends a

Loop Data message out on the link and waits for a response. The passive side returns

a Looped Data message if the message is looped at a computer. If the message is
looped at an unintelligent point such as a loopback plug, modem, or hard-wired driver,
the passive side returns the Loop Data message. The Looped Data message prevents
infinite loops between intelligent processors.

MOP is not used for link testing on an Ethernet data link.

Instead, the Ethernet

standard loop protocol is used.

7.4.2.4 System Control
MOQOP can be used to control remote, possibly unattended, systems through what can

be thought of as a virtual console. The remote system can simply be restarted, using
the Boot message. Using the Reserve Console message, an executor
can take control
of a remote system console and proceed through a complete command dialog carried in

the Console Command and Poll messages sent by the executor and the resulting
Console Response and Acknowledge messages returned by the target. The session is
ended by the executor sending a Release Console message or, in case the executor

fails, a lack of executor messages within a time-out period at the target.

Network Management

7-15

Glossary
access control

Screening inbound connect requests and verifying them against a local system account file.
-

Access control is an optional Session Control function.

active side

With regard to MOP loopback tests, the node that controls a test.

adjacency
A (circuit, node) pair. An Ethernet with n attached nodes represents n-1 adjacencies to a
/,"

Router on that Ethernet.

On DDCMP and X.25 circuits, adjacencies and circuits are in

“one-to-one correspondence, since these circuits interconnect pairs of nodes.

- adjacentnode

A node removed from another node by a single physical line.
aged packet

A packet that has exceeded the maximum number of visits.

~ASCII
American Standard Code for Information Interchange. This is a seven-bit-plus-parity code
established by the American National Standards Institute to achieve
between data services.

compatibility
|

asynchronousline
A physical line on which the time intervals between transmitted characters may be of
//’

unequal length.

Bit and byte framing are provided by the Physical Link layer on an

asynchronous line, while message framing is provided by the Data Link layer.

back-off

The Ethernet Data Link procedure which ensures stable behavior on overloaded Ethernets
by delaying retransmissions to reduce the load on the channel.

bandwidth

The range of frequencies assignéd to a channel; the difference, expressed in Hertz, between
the highest and lowest frequencies of a band. The higher the bandwidth, the more the data

throughput.

binary synchronous protocol
A data link protocol that uses a defined set of control characters and control character

sequences for synchronized transmission of binary coded data between stations in a data
communications system.

| Glossary-1

A distributed channel allocation procedure in which every station can receive all other

stations’ transmissions. Each station awaits an idle channel before transmitting, and each
station can detect overlapping transmissions by other stations.

carT

The International Telegraph and Telephone Consultative Committee, the technical

committee of the International Telecommunications Union (ITU), responsible for the
development of recommendatmns regardmg telecommumcatmns including data
communications.

channel
-

The data path joining two or more statlons 1nc1ud1ng the communlcatmns control
capablhty of the assomated statlons

characteristics

Parameters that are generally static values in volatile memory or permanent values in a

permanent data base. Characteristic are a Network Management information type.
circuit

A logical connection between protocol modules at the Data Link layer. There is one circuit
for each DDCMP point-to-point line and one circuit for each tributary station on a DDCMP
multi-point line. There is one circuit for each X.25 permanent or switched virtual circuit.
There is one circuit for each protocol type on an Ethernet.

circuit cost
A positive integer value associated with using a circuit. Messages are routed along the
path between two nodes with the smallest total cost.

client
A module that requests service of another module.

client layer
- Alayer, typically the n+1 layer, which acts as the client of the n layer.

collision

Overlapping transmissions by two or more stations on an Ethernet. The Etherent Physical

'Link layer detects colhsmns and the Ethernet Data Link 1ayer retransmlts affected data
afterarandom interval.

command node
The node where an NCP command originates.

~congestion

The condition that arises when there are too many packets to be queued.

Glossary-2

carrier-sense multiple access with collision detection (CSMA/CD)

congestion control

- The Routing component that manages buffers by limiting the maximum number of packets
on a queue for a line. Also called transmit management.

control statlon

To the Data Lmk layer protocol a station. that has overall respons1b111ty for orderly
operation of a circuit.

controller
The control hardware for a line. For a multiple line controller device the controller is
responsible for one or more units.
The controller identification is part of a line
identification.

cost

See circuit cost.

CSMA/CD

See carrier-sense multiple access with collision detection. |
data flow
.

The movement of data from a source Session Control to a destination Sess1on Control. NSP
transforms data from session Control transmit buffers to a network form before sending it

across a logical link. NSP retransforms the data at the destination from its network form
to its receive buffer form. Data flows in both directions (full-duplex) on a logical link.

data communications equipment(DCE)

The equipment that provides the functions required to establish, maintain, and terminate
a connection between DTESs using a physical circuit or a virtual circuit.

data link

A logical connectlon between two stations on the same c1rcu1t In the case of a mult1p01nt
link there can be multlple circuits.
|

data termnal equnpment (DTE)
The equ1pment typically a computer system or termmal comprising a data source and
sink connected to common carrier communication fac111tles

data transparency

The capability of receiving, without misinterpretation, data containing bit patterns that
resemble protocol control characters.

datagram

A unit of data passed between the Routing Layer and the End C.ommunication layer. When
a route header is added, it becomes a packet.

Glossary-3

\\_}/

DCE

See data communications equipment
designated router

The Router on an Ethernet chosen to perform additional dutles such as 1nform1ng the

endnodes on the Ethernet of the existence and identify of the Ethernet Routers. The Router
chosenis the one with highest priority, with highest station address breaking ties.

down-lineload

~ To send a copy of a system image or other file over a line to the memory of a target node.
DTE
See data terminal equipment.

EIA
Electronic Industries Association. A standards organization spec1a11z1ngin the electrical

and functlonal charactenstlcs of 1nterface equipment.

end node
A topological description of a nonrouting node. Since a nonrouting node cannot perform
~

route-through and supports only a single active line, it must be an end node. However, it is

also p0351ble for a routlng node w1th a smgle hne to be an end node |

end user module |
A module that runs in a user process of a network node and commumcates with Sessmn

Control to obtain logical link service.

error control
The protocol function that ensures the reliable delivery data.

|
It typically consists of

sequencing, acknowledgment, and retransmission mechanisms.

ethernet

A local area network using a Carrier-Sense Multiple Access with Collision Detect
(CSMA/CD) scheme to arbitrate the use of a 10 megabit per second baseband coaxial cable.

events
Occurrences that are logged for recording by Network Management.

executor node

The node in which the active Local Network Management Function is running (that is, the

\‘\.f'//

node actually executing a Network Management command).

Glossary-4

flow control

The protocol function that ’coordinates the flow of data between two protocol modules to
ensure that data is not lost, to prevent buffer deadlock, and to minimize communication
overhead.
|

frame
A Data Link layer message as sent or received by an Ethernet data link module or an X.25

Frame level module.

frame level
Level 2 of the CCITT X.25 recommendation which defines the link access procedure for

data exchange over the link between the DTE and the DCE.

framing
The Physical or Data Link layer component that synchronizes data at the bit, byte, and
message level.

full-duplex channel
A channel that provides concurrent communication in both directions (to and from a

station).

full routing node

An implementation of the DNA Routing layer which contains the full complement of
-

Routing components. A full routing node performs route-through functions.

gateway
A module or set of modules which transforms the conventions of one network into the
conventions of another.

half-duplexchannel

A channel that permits two-way communication, but in only one direction at any instant.
hierarchical network

A computer network in which processing control functions are performed at several levels
by computers specially suited for the functions performed, for example, in a factory or
laboratory automation.

hop

To the Routing layer, the logical distance between two adjacent nodes in a network.
hostnhode
A node that provides services for another node.

Glossary-5

interface

The relationship between different modules that are usually in the same node.

intra-Ethernet packet
A packet forwarded by a Router over an Ethernet to a destination end node which 1nd1cates
that the source of the packetis on the same Ethernet

»ISO reference model
The

International

Standards

Organization

Interconnection, ISO draft proposal DP7498.

Reference

Model for

Open

System

A proposed international standard for

network architectures which defines a seven layer model, specifying services and protocols
for each layer.

jam

|
A bit sequence transmitted by an Ethernet Data Link module upon detectmg a collision to

ensure that all affected stations detect the collision.

leased line
A nonswitched circuit leased from a public utility company (common carrier) for exclusive
use.

line
A physical path which provides direct communication among.some number of stations.

link level loopback
Testing a specific data link by sending a message directly to the data link layer and over a
- circuit or line to a device that returns the message to the source.

link management
The DDCMP component that controls transmission and reception on links connected to two

or more transmitters and/or receivers in a given direction. Also, the Ethernet data link
component whichis responsible for channel allocation (collision avoidance) and contention
resolution (collision handling).

logging

- Recording information from an occurrence that has potential significance in the operation
and/or maintenance of a network in a potential permanent form where it can be accessed by
persons and/or programs to aid them in making real-time or long-term decisions.

logging sink node
A node to which logging information is directed.

logical channel
An association between an X.25 DTE and its DCE for a given virtual circuit.

Glossary-6

logical link

'A virtual circuit between two end user processes in the same node or in separate nodes.
Session Control acts as an interface between an end user process requiring logical link
service and the End Communication layer which actually creates, maintains, and destroys
logical links.

loop node

A special name for a node that is associated with an adjacency for loop testing purposes.
|

The NCP SET NODE CIRCUIT command sets the loopback node name.

master station

A station that has control of a channel at a given instant, for the purpose of sending data

- messages to a slave station (whether or not it actually does).
maximum cost

An operator-controllable Routing Layer parameter that defines the point where the |
routing decision algorithm in a node declares another node unreachable because the cost of
the least costly path to the other nodeis excessive. For correct operation this parameter

must not be less than the maximum path cost of the network.

maximum hops
An operator-controllable Routing Layer parameter that defines the point where the
routing decision algorithmin a node declares another node unreachable because the length
of the shortest path between the two nodes is too long. For correct operation this

parameter must not be less than the network dlameter

maximum path cost

'The routing cost between the two nodes
of the network having the greatest routing»coSt,

where routing cost is the cost of the least cost path between a given pair of nodes. In Figure
3-1, the maximum path cost is 9.

maximum path length

The routing distance between the two nodes of the network having the greatest routing

distance, where routing distanceis the length of the least cost path between a glven pa1r of
nodes. In Flgure 3-1, the max1mum path cost is 9.

maximum visits

An operator-controllable Routing Layer parameter that defines the point where the packet

lifetime control algorithm discards a packet which has traversed too many nodes. For
correct operation, this parameter must not be less than twice the maximum path length of
the network.

message exchange
The DDCMP component that transfers data correctly and in sequence over a link.

Glossary-7

monitor

A logging sink thatis to receive a machme readable record of events for possible real-time
-

dec1s1on—mak1ng

multldrop line
See multipoint line

multiple line controller

A controller that can manage more than one unit. (DIGITAL multiple line controllers are
also called multiplexers.)

multlplex
To simultaneously transmit or receive two or more data streams on a single channel.

multlpomtlme
" A line connecting more than two statlons where one statlon is respons1ble for channel

control. Also referred to as multldrop

network

A collection of nodes interconnected by lines.
network diameter
The reachablhty d1stance between the two nodes of the network having the greatest
reachability distance, where reachability distance is the length of the shortest path
between a given pair of nodes. In Figure 3-1 the network diameteris 3.

node

An implementation ‘of a compyuter system that supports the Routing Layer, the End
Communication layer and Session Control layer. Each nod-_e has a unique node address.

- node address
- The unique numeric identification of a specific node.

" node level loopback
Testing a logical link using messages that flow with normal data traffic through the

Session Control, End Communication, and Routing layers within one node or from one
‘node to another and back. In some cases node level loopback involves using a loopback
‘node name associated with a particular circuit.

node name
An optional alphanumeric identification associated with a node address in a strict one-to-

one mapping. No name may be used more than once in a node.
contain at least one alphabetic character.

Glossary-8

The node name must

node name mapping table

A table that defines the correspondence between node names and node addresses or
adjacencies.

Session Control uses the table to identify destination nodes for outgoing

connect requests and source nodes for incoming connect requests.

nonrouting node
A Phase III or Phase IV DECnet node that contains a subset of routing modules and can

forward and receive packets but not perform route-through. It is connected to the network
by a single active circuit.

object type

A numeric value that may be used for process or task addressing by DECnet processes
instead of a process name.

An object type is normally used to identify a generic service,

such as file access.

oSl
See ISO reference model.

Other Data

The NSP Data Reqhest Interrupt Reqnest and Interrupt messages. These are all the NSP
data messages other than Data Segment.

Because all Other Data messages move in the

~same data subchannel itis somet1mes useful to group them together

packet
A unit of data to be routed from a source node to a destination node. When stripped of its
route header and passed to the End Commumcatlon Layer it becomes a datagram

Level 3 of the CCITT X.25 recommendation which defines the packet format and control

procedures for the exchange of packets.

' \,

‘&x—.—»/

£!

packetlevel

packet lifetime control
The Routing component that monitors adjacencies to detect if an adjacency has gone down,
and prevents excessive looping of packets by discarding packets that have exceeded the
maximum visit limit.

packet switching
See route-through
parallel data transmission
A data communication techniquein whlch more than one code element (for example b1t) of

each byteis sent or received simultaneously.

Glossary-9

parameters

DNA values to which Network Management has access for controlling and monitoring
purposes.

passive side

With regard to MOP loopback tests, the node that loops back the test messages.

path

|
A possible route for a packet from source node to destination node. This can be a sequence
|

of connected nodes between the two nodes.

path cost

The sum of the circuit costs along a path between two nodes.
path length

The number of hops along a path between two nodes.

peer protocol

in the same layer.
A protocol for communication between modules
|

permanent virtual circuit

A virtual circuit between two V.25 DTEs that is always established. A logical channel is
permanently allocated at each DTE/DCE interface to a permanent virtual circuit.
|

physical link

An individually hardware addressable communication path.
piggybacking

Sending an acknowledgment within a returned data or control message.
pipelining

Sending messages without waiting for individual acknowledgment of each successive

message.

point-to-point circuit
A link connecting only two stations.

protocol

A setof messages with specific formats and rules for exchanging the messages.
raw event
A logging event as recorded by the source process, incomplete in terms of total information
required.

Glossary-10

\-a—"/

reachablenode
- A node to which a routing node believes it can direct a packet.
reassembly
The placing of multiple, received data segments by NSP into a single Session Control
~

receive buffer.

remote node
To one node, any other network node.
request count
This term has two definitionsin NSP: (1) Variables that NSP uses to determine when to

send data, and (2) values sent in Link Service (Data Request and Interrupt Request)
messages.

The flow control mechanism adds the request counts received in Data Request and
Interrupt Request messages to the request counts it maintains to determine when to send
_
-

data.

retransmission

The resending of messages that have not been acknowledged within a certam period of
time. Thisis part of aprotocol’s error control mechanisms.

RMS
Record Management Services. This file system will be used on all major DIGITAL systems

except where space is limited (for example, RT-11). In addition to access modes provided
by
‘previous file systems. RMS provides random access for direct and indexed files and ISAM.
router
See full routing node.

route-through
The directing of packets from source nodes to destination nodes by one or more

intermediary nodes. Routing nodes permit route-through. Also called packet switching.

routing

|
D1rect1ng data message packets from source nodes to destination nodes.

routmg node
"

See full routmg node

segment
The data carried in a Data Segment message. NSP divides the data from Session Control
transmit buffers into numbered segments for transmission over logical links.

Glossary-11

segmentation

The division of normal data from Session Control transmit buffers into numbered segments
for transmission over logical links.

A module or set of modules in a layer that perform a well defined service, such as remote
file access or gateway communication on behalf of another module.

server system

A node which contains one or more servers. A server system is often dedicated to server

server

functions.

sink node
.<

A node which receives and records Network Management events.
slave station

A tributary station that can send data only when polled or requested to by a master. control
station.

~ star topology

A network configurationin whlch one central nodeis connected to more than one adjacent
end node. A star can be a subset of a larger network.
|

station

With regard to the Data Link layer protocol, a termination on a data link. A station is a
combination of the physical link (communication hardware) and the data 11nk protocol
|

implementation.

station address
An address assigned at the data link level to a station.
status

Dynamic information relating toa network such as a line state. A Network Management
information type.

subchannel

A logical communication path within a loglcal link that handles a defined category of NSP
data messages. Because Data Segment messages are handled differently from Other Data
messages, the two types of messages can be thought of as travelling in two different
subchannels.

switched virtual circuit

A temporary association between twoX.25 DTEs.

Glossary-12

synchronousline
-

A physical line on which data characters and bits are transmitted at a fixed rate with

transmitter and receiver synchronized. Bit framing is provided by the Physical Link layer
on a synchronous line, while byte and message framing are provided by the Data Link

layer.

target node

The node that receives a memory image during a down-line load, generates an up-line

dump, or loops back a test message.

topology
The physical arrangement and relationships of interconnected nodes and lines in a
network. A legal topology satisfies the requirements of the Routing Layer specification.

transparent data

Binary data transmitted with the recognition of most control éharacters suppressed. For

example,

DDCMP provides data transparency because it can receive, without
misinterpretation, data containing bit patterns that resemble DDCMP control characters.

tributary station
A station on a multipoint line that is not a control station.

unit

The hardware controlling one circuit on a multiplé line controller. A unit, a controller, and
associated Data Link modules form a station. See Figure G-1.

unreachable node

A node to which a routing node has determined that the cost of the least costly path exceeds
the maximum cost of the network, or the length of the least costly path exceeds the

maximum hops of the network.

up-linedump

To send a copy of a target node’s memory image up a line to a file at the host node.
user

A person or module which requests service from a layer. Client is the preferred terminology

when referring to modules.

virtual call
See switched virtual circuit.

virtual circuit
A connection between a data source and a sink in a network. They may be realized by
different circuit configurations. Virtual circuits typically provide guaranteed delivery and
sequentiality of client data.

Glossary-13

wild card

With regard to DAP, an asterisk (*) that replaces an element in a file specification. The
asterisk specifies all known items in the range indicated by its position. For example,
FILE.*;* specifies all known type, and versions of all files named FILE.
window

A range of packets authorized for transmission across an X.25 DTE/DCE interface.

~ Glossary-14

DECnet Architecture

General Description
AA-N149A-TC

READER’S COMMENTS

NOTE: This form is for document comments only. DIGITAL will use comments submitted on this fofin_ét the
company’s discretion. If you require a written reply and are eligible to receive one under Software
Performance Report (SPR) service, submit your comments on an SPR form.

,/;

Did you find this manual understandable, usable, and well-organized? Please make suggestions for improvement.

\Wl

| Did ybu find errors in this manual? If so, specify the error and the page number.

Please indicate the type of user/reader that you most nearly represent.

Assembly language programmer

Higher-level language programmer

Occasional programmer (experienced)

User with little programming experience

Student programmer

Other (please specify)

Name

Date

Organization

)

Street
City

State

Zip Code
or

Country

— == — ==[)o Not Téar - Fold Here and Tape — — — — — — | ———

— e —

Necessary

if Mailed in the

United States

dlilg 1l

No Postage

BUSINESS REPLY MAIL

FIRST CLASS PERMIT NO.33 MAYNARD MASS.

i
|

SOFTWARE DOCUMENTATION

\\.

POSTAGE WILL BE PAID
BY ADDRESSEE

1925 ANDOVER STREET TW/EOQ7

—

Cut Along Dotted Line

—

— — — —

Do Not Tear - Fold Here and Tape

— — —

TEWKSBURY, MASSACHUSETTS 01876