Digital PDFs

MISC-0001-PDP-11

2000

43 pages

Original

2.1MB

Document:	DECNET TECHNICAL SUMMARY PDP-11
Order Number:	MISC-0001-PDP-11
Revision:	0
Pages:	43
Original Filename:

OCR Text

DECNET
TECHNICAL. SUMMARY
PDP-11

Distributed
Data Processing
Group
~DmDDmD

DECNET
TECHNICAL. SUMMARY

PDP-11

PRINTED IN U.S.A.

TABLE OF CONTENTS

PART I-DECNET

PART II-COMMUNICATIONS HARDWARE

CHAPTER 1-INTRODUCTION

CHAPTER 6-HARDWARE COMPONENTS

1.1
12
1.:3
1.4

DECnet and Networks .............................. ..........
DECnet-General Description ............................
DECnet Functions ..............................................
DECnet Building Blocks ....................................

3
3
4
4

CHAPTER 2-DECNET FEATURES
2.1
2.~~

2.3
2.4

DECnet Features and Benefits ..........................
Programmable Calls and Other User Interfaces
Down-Line Loading and Remote Program
Development and Storage .................. ................
Special Considerations ......................................

7
8
8
10

CHAPTER 3-DIGITAL NETWORK ARCHITECTURE
3.1
3.~~

3'::~

3.4
3.5

Structured Design ..............................................
Layered Protocols ..............................................
DDCMP ................................................................
NSP ......................................................................
DAP ......................................................................

13
13
14
15
16

C.iAPTER 4-S0FTWARE SYSTEM COMPONENTS
4.11
4.~~

4.:3

RSX11 Family......................................................
RT11 Operating System ......................................
RSTS/E Timesharing System ............................

17
18
18

CHAPTER 5-NETWORK BUILDING BLOCKS
5.1
5.~~

5.~~

5.4

Network Building Blocks ....................................
DECcomm 600 ....................................................
Remote Computer Systems ................................
IBM Front End ....................................................

21
21
22
23

6.1
6.2
6.3
6.4

Processors ............... ........ ......................... ..........
Asynchronous Links ............................................
PDP-11 Asynchronous Links ..............................
Configuring Systems with Asynchronous
Interfaces ............................................................
6.5 Synchronous Links ............................................
6.6 PDP-11 Synchronous Interfaces ........................
6.7 Configuring Systems with Synchronous
Interfaces ............................................................
6.8 Synchronous Link Example ................................
6.9 Parallel Links ......................................................
6.10 Distance Considerations ....................................
6.11 Special Adapters ..... ...........................................
6.12 DECnet Interconnections ....................................

27
28
28
29
31
31
32
34
34
34
35
35

CHAPTER 7-MODEM SELECTION GUIDE
7.1
7.2
7.3
7.4

Asynchronous Modems ......................................
Synchronous Modems ........................................
Synchronous (Wideband) Modems) ..................
Modem Options ..................................................

37
37
37
37

APPENDIX A-SUMMARY TABLES
A 1-DECnet Interconnection Facilities ......................
A2-Supported DDCMP Modes ..................................
A3-Supported NSP Features ....................................
A4-Supported DAP Features .. ............. ...... ..... ..........
A5-DECnet Functions By Host Environment Pairs..
A6-DECnet Interconnectability Matrix ......................
A7-DECnet Product Reference Table ......................

39
39
39
40
40
41
41

PART 1- DEGNET

CHAPTER 1
INTRODUCTION
The intent of this manual is to help you understand
"DECnet" (DIGITAL's computer network products) as it
relates to the PDP-11. There are sections describing
DECnet implementation, sections to assist in the
selection of PDP-11 hardware and software, and
sections describing network-related products.
Part I is devoted to DECnet implementation and DECnet
supported products.
Part II is devoted to DEC products that can be used in a
network implementation. These products mayor may not
be currently supported by DECnet. However, they can be
used in the initial phases.
1.1 DECNET AND NETWORKS
The term network covers a spectrum of computer configurations. "Terminal Network" sometimes is used to
describe a central processor, its peripherals, the
terminals that use it, and the links that connect the
central processor that is linked to stations having card
readers and line printers. People use such networks to
submit batch jobs via the card reader, and obtain the
results on the line printer. Examples: IBM2780
networks: CDC UT -200.
DECnet falls into another category of Computer
Networks. Its salient features are:
1. More than one "node" (a CPU and its associated
peripherals) is in the net, being connected via
communications links.
2. E:ach node is potentially capable of performing
application oriented tasks. This distinguishes
DECnets from terminal and remote job entry
networks.
3. No master or "host" CPU is required.
This last category can be further subdivided between
closely-coupled and loosely-coupled networks. Closelycoupled networks are those with nodes linked via
internal data busses or as a unit; changes in architecture
are not easily made. Furthermore, closely-coupled
networks usually require all nodes to be in close
proximity.
In contrast, loosely-coupled networks such as DECnet
offer far greater flexibility in terms of network architecture,
functionality and geographical location of the nodes.
Network communication can be highly integrated into a
specialized system that acts as a unit, or it can be limited
to occasional transfers of data -files, leavi ng the nodes
quite independent. Since DECnet nodes interact over
communications links, the nodes can be in the same
room, or halfway around the world.
DECnet is a series of software packages that are fully
integrated into DIGITAL's major operating systems and
allow easy intercomputer communication over com-

munications links. DECnet is a group of tools that may
be used to create networks useful in a variety of
applications.
1.2

DECNET-GENERAL DESCRIPTION

DECnet is the collective name for the set of software
products which extend various DIGITAL operating
systems so they may be interconnected with each other
to form computer networks. The DECnet user can configure a variety of networks, satisfying a variety of
constraints, by choosing the appropriate CPU's, line
interfaces (and speeds) and operating systems software.
Such networks typically fall into one of three classes:
Communications Networks: These networks exist to
move data from one physical location to another, often
distant, location. The data may be file-oriented (as is
often the case for remote job entry systems) or lineoriented (as occurs with the concentration of interactive
terminal data). Interfaces to common carriers, using both
switched and leased-line facilities, are normally a part
of such networks. Such networks are often characterized
by the concentration of all user applications programs
and data bases on one or two large "Host Systems" in
the network.
There are a variety of reasons for implementing communications oriented networks. One major motivation
for such networks is to reduce communication line costs.
Common elements of such a network are data
concentrators (which concentrate a number of lines into
a single line back to the main site) and communication
front-end preprocessors (which relieve a large computer
system of chores associated with communications and
line usage). Another type of Communications Network
is the type which is aimed at faster movement of information and reduction of paper work. For instance, credit
verifications and cash transactions are being handled
more and more by on-line terminals and computer-tocomputer networks.
Resource-Sharing Networks: These networks exist to
permit the sharing of expensive computer resources
among several computer systems. Shared resources can
include powerful central-processors, mass storage
devices, unit record equipment, and other peripherals.
The resource to be shared may be a logical, rather than
physical, entity, as in a data base which is stored
centrally and made available to other machines in the
network. Such networks are often characterized by the
concentration of large peripherals, large data bases, and
large programs on one or two "Host Systems" in the
network, while "Satellite Systems" have smaller
peripherals and programs.
Distributed Computing Networks: These networks exist
to coordinate the activities of several independent

computing systems toward some larger goal. Networks
of this nature may have specific geometries (star, ring,
hierarchy), but often have no regular arrangement of
links and nodes. Such networks are usually built to put
computational power close to the users of such power.
Distributed Computing networks are usually characterized by a multiplicity of comparably sized computers
and applications programs and data bases distributed
throughout the network.

Down-Line Program Loading: Programs to be executed
on other systems in the network may be stored on the
local system, and loaded on request into other systems,
under the joint control of the operating systems at both
ends of the network. This and the preceding feature
simplify the operation of network systems without mass
storage devices, by allowing such systems to use
remote mass storage devices in a convenient and
straightforward manner.

In ordBr to satisfy these widely varying constraints,
DECnBt allows the user to build networks from a range
of system and communication components.

Down-Line Program Commands: Programs running on
one system in the network may cause program commands to be executed at remote systems in the
network. This feature allows remote pro~lrams to be
started and stopped.

DECnet includes a set of network protocols, each of
which is designed to fulfill specific functions within the
netwo 'k. Collectively, these protocols are known as the
Digital Network Architecture, or DNA. The major
protocols, and their functions are:

Inter-System File Transfer: This factlity allows an entire
data file to be moved between systems, at either program or operator request.

Digital Data Communications Message Protocol
(DDCMP): DDCMP handles the link traffic control and
error recovery within DECnet. DDCMP has been
designed to operate over full and half-duplex facilities,
using synchronous, asynchronous, and parallel facilities.

Cross-System Support: This facility allows program
development actitivities to be performed on a system
different from the one where the programs will be
executed.

Netwolrk Services Protocol (NSP): NSP handles network
mana~lement functions within DECnet, including the
routin(~ of messages within any given system.

High-Level Language Interface: This facility allows
programs written in compiler languages (FORTRAN,
COBOL, BASIC) to access some or all of the network
facilities.

Data J~ccess Protocol (DAP): The Data Access Protocol
enables programs on one node of the network to utilize
the I/O calls into the DAP standard, and vice-versa.
DAP thus allows remote file access, including OPEN,
READ, WRITE, CLOSE and DELETE for sequential and
random files, and remote device access, for unit record

A goal for the set of DECnet products has been to provide as general an interconnection mechanism between
specific products as possible, limited only by the technology and cost considerations which constrain each
individual member of DECnet. These latter constraints
make totally general interconnectability impractical.

devicE~s.

DECnet permits communications between various
Digital Equipment Corporation processors (DECsystem 10's, PDP-S's and PDP-11 's) and various
operating systems. This manual is mainly devoted to
implementation of communications networks using
PDP~ 11 systems. PDP-11 operating systems which will
support DECNET include:

1.3 DI:CNET FUNCTIONS
Digital Network Architecture (DNA), implemented across
a wide range of operating systems and hardware architectums, enables users to build a variety of networks, as
previously described. Such networks have some common
attributes, and individual systems in the network may
have certain system-specific attributes. The common
attributes include:

RSX11M
RSX11D
RSX11 S

Inter-program Communication: Programs on one system
can exchange data with programs on other systems in
a real··time fashion.

RSTS/E
RT11
lAS

1.4 NETWORK BUILDING BLOCKS
DECnet allows the implementation of advanced net ..
work capabilities-for instance such network functions
as: program-to-program communications, task-to-task
communications, and device-to-device communications.
Many network users may require less advanced links and
capabilities or may prefer to move toward a full DECnet
implementation in a phase-by-phase approach. For this
class of users DIGITAL offers a series of network
"building block" systems which can be used in a
"Phase I" implementation and can later be upgraded
into a more comprehensive DECnet network. These
building blocks are summarized below and described
in more detail in Chapter 5.

Inter-!:iystem Resource Sharing: Programs on one
system can utilize files and devices physically attached
to othE3r systems in the network.
Additionally, many DECnet systems support other
featums which are useful in the network environment.
These include:
Down .. Line System Loading: Initial core images for
other systems in the network may be stored on the
local system, and loaded on request into other systems
in the network. The remote systems usually require the
use of a network bootstrap loader, implemented in
read-only memory.

Building Block Summary

PDP-11 OPERATING SYSTEMS
RSX-11 D

Powerful multi-task, operating system
with multi-terminal support.

RSX-11M

Midsize real-time operating system.

RSX-11 S

Small satellite or standalone "core
only" system.

RT11

Single user operating system.

RSTS/E

Multi-User BASIC Timesharing System
SYSTEMS WITH FACILITIES FOR
COMMUNICATING WITH· IBM

RSX11 D/2780

RSX11 D system with 2780-type
communications.

RSX-11 M/2780

RSX-11 M system with 2780-type
communications.

RSTS/E-2780

RSTS/E time-sharing system with
2780-type communications.

RT11/2780

RT11 system with 2780-type
communications.
BUILDING BLOCK HARDWARE

PDP-11 Family Processors
DX11 PDP-11-to-IBM channel interface
DU11, DQ11, DV11 synchronous interfaces
DL 11, DJ11, DH11 asynchronous interfaces
In addition to the above building block hardware and
software, DIGITAL's Special Systems Group (CSS) can
develop special hardware and software. For instance
CSS has developed special switches for connecting
busses for multiple PDP-11 CPU configurations,
PDP-11 /CDC channel interfaces, and HASP workstation
emulation software.

CHAPTER 2
DECNET
2.1 DEeNET FEATURES AND BENEFITS
The following table summarizes the capabilities of DECNET nodes.

Benefits
Total operating system acts as a unit; network user
programs have all standard systems facilities available
to them.

Feature

FULLY-INTEGRATED SYSTEM SOFTWARE.

UPWARD COMPATIBLE IMPLEMENTATIONS on
PDP-8's, -11's and DECsystem-10.

Gives a tremendous range of performance that allows
easy configuration of the optimal network for each
application. The larger implementations on the PDP~ 11
and DECsystem~ 10 have added capabilities in terms of
number of devices, number of users, and flexibility.
However, within the limits of the types of links
supported, any network node may be connected to any
other one with no modification to system software
except possibly re~SYSGEN or re-TASK~BUILD.

DEVICE-SHARING: The ability to
access remote peripherals.

This allows (1) uSers of terminals on one system to
access facilities and terminals on another system, (2)
programs to .share low usage peripherals among systems
to reduce costs, and (3) system managers to create
development systems for access by small, execute-only
computers.

FIL.E SHARING: The ability for programs and terminal

This allows creation of (1) distributed data base support
systems, where sensitive data may be stored locally and
given Hmited external access, and (?) data acquisition
systems, where the file system is remote and the data
gathering computers may be in a "toxic" environment.

USEHS to access remote files, provided they satisfy

security requirements.

DOWN-LINE LOADING: The ability to transmit an
operating system to a remote computer and have it run.
there.

This allows the construction of systems where computers
need not have mass storage capability (disks or
tapes), but are able to run a variety of applications by
requesting programs from other computers in the
network.
(SEE DOWN-LINE LOADING).

REMOTE PROGRAM DEVELOPMENT AND STORAGE:
The ability to create and debug a program on one
computer and run it on another.

Applications programmers who need to code network
data exchanges need know very little about communications.

EASY-TO-LEARN PROGRAM INTERFACE for MACRO
and FORTRAN programs.

Allows flexibility in choosing the most cost-effective link
for each application. When a node supports more than
one type of link, main memory requirements are reduced,
because the same protocol code can support all links
on the node.

INDEPENDENCE OF LINK CHARACTERISTICS, allowing
use of same protocol on asynchronous or synchronous
serial links and parallel links.

Gives added flexibility in network design, and may help
in reducing common carrier tariffs.

FIXED-PATH ROUTING: The ability to send messages
from one node to another through one or more
intermediate nodes.

Users can virtually ignore the problems of correct sequencing and error detection and correction previously
associated with data communications.

AUTOMATIC ERROR CORRECTION:

Allows users to take advantage of better price/performance of high-speed links.

MUL TIPLEXING/DEMULTIPLEXING: The ability to carry

on several, distinct conversations over a single
communications link.

Unlike communications software using earlier link protocols, this feature allows total utilization of a full-duplex
link, and increased throughput.

TRUE FULL-DUPLEX OPERATION: the ability to simul~
taneously send data in both directions on a full-duplex
link.

Incases throughput and allows efficient use of links, such
as those associated with satellite transmissions.

MULTIPLE MESSAGE ACKNOWLEDGMENT: The ability
to send more than one message before the first message
has been acknowleged as correctly received.

Aids network maintenance and testing applications
programs.

NETWORK EXERCISER PACKAGE:

2.2 PROGRAMMABLE CALLS AND OTHER USER
INTEI=lFACES

satellite operating systems) module for final compilation
and/or execution. The load module may be stored at the
host for future runs.

Users can employ the DECnet features at three levels:

Operator-and Program-callable File Transfer Functions

1. Down-line loading of programs and/or whole systems
under program or operator control.

The file transfer functions allow:

2. Fil'9 creation, transfer access and deletion via calls

1. Local/remote, remote/local, and remote/remote
transfers of whole files. This means the location of all
files in the network can be controlled from one place.
2. Local/remote and remote/local, sequential transfer
of records (programs only). This means (for example)
that small data collection nodes need only enough
storage to handle one record at a time .
3. File allocation, creation, and deletion. No interaction
with a remote operator is required for standard file
operations.
4. Directory creation and deletion. This completes the
tools required for creating a distributed data base
management system.
5. Transmission of files containing commands for remote
monitors,

to subroutines.
3. Inter-program communications via calls to system
primitives.
Example:

Local
Task Code

. Remote
Task Code

CALL NTCON

This call asks to start the This task will be awakened
"conversation" by re- when other task issues call
questing creation of a log- to NTCON.
iCed link."
CALL NTXMT

CALL NTRCV

This sends the first piece
of the conversations.

No communication is possible until receive has been
issued.

CALL NTRCV

CALL NTXMT

System Primitives
The system primitives allow:
1. Creation and destruction of "logical Ilinks," DIGITAL's
term for distinct conversations betw6'en programs.
2. Transmission and reception of data messages between programs. There is no restriction on data
exchanged, except that the recommended maximum
for a single message is 528 bytes. Provided extra
memory is available, longer messages can be transmitted.
3. Interrupt message facility. This facilitates synchronization and allows checkpointing (rollin/rollout) of network programs.

This gets the answer.
CP.LL NTDIS
This ends the conversation. The system will inform
this task.

2.3 OIOWN-LiNE LOADING AND REMOTE PROGRAM
DEVE:LOPMENT AND STORAGE
Down-line loading allows quick re-booting in nodes without mass storage devices. Remote program development
is the analogous capability at the program level. With
this capability, customers can create memory-only nodes
for reduced cost, or for placement in "toxic" environments. At the memory-only node, down-line load is implemented with a read-only memory bootstrap, which
operates on special boot messages sent from the host.
The host computer has a resident system program to
format the messages and operate the boot-mode
protocol.

All non-network functions present on the supporting
operating systems are still available when the network
support features are added. The network features are
strictly enhancements, and they are fully integrated with
the rest of the system software. No applications programs need to be re-written to run within a network node,
unless network extensions to the applications are
desired.
MACRO assembler and FORTRAN interfaces are available in all supporting operating systems . Coding network
data transfers is almost as simple as coding transfers
to and from a disk.

Remote program development allows programmers to
write and debug their programs at a central node, and
ship the source or load (depending on the host and

COMPo f
TERMINALS

R5X-llD
PDP-ll

RSX-llD
PDP -11

COMPARE

Figure 2-1

Message Switch
Source Node A

Intermediate
Node B

compose messag~CALL NTRCV
chBck address
check address
of next node
of next node
on route
on route

[321,4567] specifies the user's area. COMP.F is the name
the file will be given. The characters that follow the equal
sign specify the source file. No source node name is
supplied, so the local node is assumed. DKO:[200,200]
specifies the source of the file (RK disk zero, area
200,200), and COMP.PRE is the name of the source file.

Destination
Node C

CALL NTRCV
act on
message

Remote Record Transfer

re-form message,
if needed
/
CALLNTXMT

Figure 2-2 illustrates a resource-sharing network. As in
the previous example, the communications links enhance
the capabilities of the outlying systems: they may allow
them to support fewer, less expensive peripherals, but
in essence, they are only a convenience. The same job
could be done without the links, but with greater inconvenience, greater overall cost, and less efficiency
for all concerned.

CALLNTXMT

EXAMPLES
Remote File Transfer

In Figure 2-1, an RSX11 0 system running on a PDP-11
is preparing data for a computational program that runs
on another PDP-11 based RSX11 0 system. The communications link in this case is only a convenience that
allows faster turnaround, being preferable to unloading a
magnetic tape with the formatted data and physically
transporting it to a (possibly) distant computational
center.
The communications part of the operation is relatively
easy under DECnet, because the input file may be transferred by a single command from one of the terminals.
The link is probably a dial-up line, because constant
communication is not require~.
If the file is transferred via an operator command, the
command might look like this:
PLE>PDP-11A-DP1 :[321 ,4560]COMP.F=
DKQ:[200,200]COMP.PRE

Figure 2-2

The first four characters of this command (PLE» are
the prompting characters supplied by the system program,
PLE. The next logical group is terminated by the equal
sign (PDP11A-DP1 :[321 :4567]COMP.F=), and represents the target of the file transfer. PDP11A is the name
of the node, and underscore (-) terminates the name.
DP1: specifies which device will receive the data, and

Programming required for the communications part of
the applications will vary with the way the system is used.
If data collected in the outlying systems are stored locally
and transferred as a file, no network programming is required--just an operator command. If no local storage
is available, or if some real-time coordination between

an outlying system and a task at the central computer is
requirecl, however, the system primitives must be used.
This is how the code might look:

Node A

Node B

CALL NFOPN

This subroutine opens the file.

CALL NFPUT

This subroutine stores a single logical
record.

Calculate event's
Call NTRCV
parameters
/ ( T h i S receives the
message).
•
Analyse event
•
parameters
•

CALL NFPUT

This call stores another one.

Compose message

CALL NFCLS

This subroutine closes the file.

CALL NTXMT /
(This transmits the
message).

No user code or operator intervention is required for the
central node, because the internal network system code
handles all such transfers.

•

Set ~vent flag

Suppose the process must be changed, and a different
program must be loaded into a control system. This is
how that would happen:

Note that the programming interface need not be affected
by the characteristics of the link (full or half-duplex,
synchronous or asynchronous, etc.).

SUPERVISOR NODE
Determine new program required

Comple!x Inter-program Communication

Determine name
of task required

The lowest level in a certain network may include simple
computers (with no mass storage) operating In hostile
environments, as well as disk-based computers. Such
machim~s may be used in real-time control of various
processes and other on-line functions in the factory. They
may have digital and analog acquisition devices and may
perform limit checking and control algorithms or directly
control and alter the process through the use of digital
and analog outputs.

CALL NTTLD
This subroutine transmits
the load image from storage
to the remote computer.
CALL NTCON
Request logical connection
(conversation) with task
just sent over the link.

The intermediate level computers provide supervisory
control to the assembly line drivers by changing the control parameters and disciplines which they are using.
In addition to integrating the assembly line into a singlesystem that acts to control a single large process, the
mid-level machines may also acquire data from various
manual entry stations and terminals throughout the
factory. The manual entry stations may record material
movement, machine utilization, production quantities, employee attendance, inspection results and the like. The
terminal may allow real-time inquiry from various managerial levels about order status, equipment use, up-timet
down-time, and so on.

Note that no coding is required at receiving end.
2.4 SPECIAL CONSIDERATIONS

This section is intended to show some limitations of
DECnet, and some possible alternatives.
Communications With Other Manufactulrers' Systems

DECnet specifications will be published, allowing
customers with other manufacturers' equipment to code
compatible network software. For customers who do not
wish to implement DECnet specifications, other software
products, such as IBM 2780 packages, alre available at
extra cost.

The data base, which reflects overall operation of the
factory, may be distributed between the mid-level and
top-lev '91 machines. The latter may be a large-scale
PDP-11 /70 under RSX-11 D. The data base processing
can be done by that system, or delegated to the midlevel machines, when appropriate.

Adaptive Route-Adaptive

DECnet employs routing techniques that require a
fixed structure for the network.
Reliability

In this network, the communications links are fully integrated into the solution of the problem. Applications such
as this one may not be solvable by anything but a computer network. The program interface for network communications is likely to be more complicated, only because more is required of the real-time program involved.
The actions must be coordinated across machines, so'
events must be posted, programs must be activated, and
data must be passed between nodes. These activities
are greatly simplified by the fact that RSX-11 M, and -11 D
are fully upward compatible.

DECnet provides tools to make "loosely coupled"
high-availability systems. These tools, combined with
existing DIGITAL options such as multi-port disks and
UNIBUS windows, and the high-availability experience of
our Computer Special Systems engineers make DIGITAL
a major vendor of high-availability systerns. This does
not, however, mean that DECnet provides instantaneous
reliability. Several independent aspects l11ust be
considered.

This is how a program might post an event in another
machine:

2. High-availability Techniques-The designer of a highavailability system must understand the requirements

1. Link Reliability-DECnet protocols may not be
efficient over extremely noisy links.

that redundancy imposes on data storage, file
updates, and other critical areas. It is recommended
that if such a person does not exist in-house, the
customer contact DIGITAL's Computer Special Systems division or a qualified systems house for
consultation and assistance.
3. High-reliability Acknowledgement Scheme-DECnet's
acknowledgement scheme solves the problem
of lost messages in every case except when a node
receives a message but "goes down" before
processing it. A user-level acknowledgement scheme
will ensure full reliability.
Transparency

Network I/O must be explicitly called as such. It is not
currently possible to treat I/O devices on a remote node
as lif they were local. Thus, an application program will
typically require some modification if it is to use
resources located on other nodes in the net. This
approach may appear inconvenient, but it offers many
opmating advantages. Most of the burden for I/O
processing (e.g., buffering and record blocking and
deblocking) is placed on the (presumably larger) system
at the other end of the link. Additionally, the FORTRAN
interface for network facilities allows the user to initiate
asynchronous operations, while normal (e.g., local) I/O
operations in FORTRAN are service synchronously (that
is, the user's program is delayed until the operation is
complete). Similarly, many internal operations in remote
nodes must be accomplished by a cooperating task in
the remote node. For example, in order to post an event
in a remote RSX-11 D system, the local program must
send a message to a remote program with the parameters
required to specify the event. That task must then initiate
a monitor call to post the event.
Multipoint (Multidrop) Communications

DECnet currently does not support multidrop links
(i.e., physical communications links with more than two
nodes to a link). However, the DDCMP specification
includes provisions for multi-drop implementation.

CHAPTER 3
DIGITAL NETWORK
ARCHITI=:CTURE
3.1 STRUCTURED DESIGN
Our design considers the network and its corresponding
communication facilities as a number of distinct layers.
The entire system of "layers" is called DIGITAL Network
Architecture (DNA). Within each layer the functions have
be!3n logically separated into modular units. This structured design allows for ease of subsetting and
modifications within our network implementations. At
each layer the functional modules are independent and
may be implemented as either individual tasks or subroutines within a task or system. This allows creation of
nodes with partial functional capability. The layered
approach to communications allows modification and/or
replacement of a specific protocol without disturbing
other layers in the structure.
To make the network as useful as possible there are no
specific constraints put on the functional capabilities and
their relation to the structure of the net. The physical link
protocol DIGITAL Data Communications Message Protocol (DDCMP), allows for point-to-point or multipoint.
half or full duplex, or serial (asynchronous or synchronous) or parallel links within the net. The addition of routing through capability at network nodes extends communications to hierarchical and random configurations.
These capabilities will allow hierarchical and random configurations. These capabilities will allow line cost
reductions in some situations and increased speed and
reliability through user-level alternate path routing
capability in others.

3.2 LAYERED PROTOCOLS
Data to be sent from a user program in one node to a
user program in another is enveloped within two or three
protocols and passes through various interface layers as
parameters required by these protocols are added and
subtraced around the user data. The highest layer is the
end-to-end or user intertask communications mechanism
layer. The next layer must multiplex these many user
messages into a single queue or data stream for eventual
transmission over a physical data link. This multiplexing
usually consists of the addition of header routing
information allowing demultiplexing at the receiving end.
This data stream is then passed to a link layer protocol
concerned with error detection and correct sequencing
messages being sent between adjacent nodes. This layer
communicates with a driver layer protocol that interfaces
with the hardware of the system. The levels in the system
may thus be divided as follows:
Dialogue Layer

The conversation created from
the messages sent over a logical
link into a meaningful exchange
between users.

Logical Link Layer

The multiplexing/demultiplexing
of the link message stream into

individual message streams for
each logical link.
Physical Link Layer

The interface of message packets
to/from the hardware layer,
concerned with message error
detection and recovery, message
sequencing and message synchronization over the link.

Hardware Layer

The transmission/reception of the
data bits over a physical link,
concerned with transmission
techniques (synchronous,
asynchronous, parallel), character
synchronization and modem
operation. Provides device
independent interface for upper
layers.

The dialogue layer depends entirely on the application.
It involves both messages and actions taken as a result
of the messages. From DECnet's perspective, it may
involve several distinct conversations, or "logical links."
The logical link layer creates logical links via the Network
Services Protocol (NSP). Network Services Protocol
messages and rules support creation of logical links and
transfers of user data over those links. The physical link
layer implements the link protocol, DIGITAL Data Communications Message Protocol (DDCMP), which ensures
proper sequencing of messages, detects and recovers
from errors, and controls "ownership" of half-duplex
links. The hardware iayer is handled by a device driver
and interrupt service routine. Its purpose is to exchange
data with the hardware interface, handle synchronization,
and modem control.
Within DECnet these layers have been related to
specific code function, protocols and commands.
Figure 3-1 illustrates the layered approach to network

USER-ORIENTED
FUNalONS (DAP)
NETWORK
SERVICES & FACILITIES
(NSP)
PHYSICAL LINK
MANAGEMENT
(DDCMP)

------HARDWARE
INTERFACE

Figure 3-1
Layered Approach Applied to General Network Design

design, and compares it with typical contemporary
operating system design. Layered systems provide:
USER CRE:ATES DATA

Easy debugging of layers
CI ean interfaces
Easy replacement of layers
Easy subsetting of layers
Easy integration with operating systems
Ability to move hardware/software interface line

USER TASK DATA

NETWORK SERVICES
USER TASK DATA

Figure 3-2 illustrates the way structured design can be
implemented. Figure 3-3 shows data flow through the
systern. Note how first the routing header, then the link
protoGol header are added, then the driver transmits the
message over the communications link, and finally the
headers are subtracted, in reverse order.

PHYSICAl. LINK PROTOCOL

TAS~: DATA

LINK PROTOCOL
HEADER

USER

LINK PROTOCOL
HEADER

USER TASK DATA

BLOCK
. CHECK.

SENT OVER LINK

PHYSICAL LINK PROTOCOL
USER TASK DATA

NETWORK SERVICES
USER TASK DATA

Figure 3-3
Data Flow Through the System

assembled in the proper sequence at the receiving
station. They may arrive out of order due to loss during
transmission or duplication by the error recovery
mechanism.
The numbering is initialized (usually to zero) by the start
message (one or the protocol messages). The number
field in the header of each succeeding data message is
incremented by one modulo 256 over the preceding
data message. That is, the numbers range from zero to
255 and then wrap around.

Figure 3-2
Structured Design

The receiving station acknowledges correct, in-sequence
messages by returning the message number either in the
response field of data messages transmitted in the
opposite direction or in a separate acknowledgment
message (ACK).

3.3 DDCMP, DIGITAL DATA COMMUNICATIONS
MESSAGE PROTOCOL

DIGITAL has produced a data link protocol that operates
on existing hardware and is implementable on many
opera.ting systems.

Each message does not require a separate acknowledgment. An acknowledgment of one message implies
acknowledgment of all preceding ones. This scheme,
coupled with capability of acknowledging messages in
the header of data messages going in the other
direction, allows for efficient use of the data link.

DDCMP, Format

Data is transmitted in variable length blocks called
messages, which consist of the data, along with the
control and error-detection information needed to check
and order this data at the receiving end. All data
message headers follow a prescribed format. Besides
data messages, there are protocol messages for
acknowledgment, initiating message exchange, and
error recovery.

Whenever an error is detected, a negative acknowledgment (NAK) message is returned to thE! transmitter. The
NAK contains the number of the last correctly received
message, providing two pieces of information: it
acknowledges all messages up through the number
given, and implies the need for retransmission for all
subsequently sent data messages. Up to 255 unacknowledged messages can be outstanding before an
acknowledgment becomes mandatory.

TherE! are two parts to a data message-the header,
which contains control information, and the data itself.
Data messages are numbered so that they may be
14

Tille Data Message Header
The data message header contains control information,
such as a count field that indicates how long the data
message is, a message number for sequencing
messages and determining loss of a message, and an
acknowledgment field that allows message acknowledgment without a separate ACK or NAK message. The
hElader ends with a 16-bit block check used for
detection of errors in the header.
The Data Field and Block Check
The data consists of a stream of bits that must number
some multiple of 8. The bit patterns in the data field
have no effect on the protocol, so any coding scheme,
character size or control characters may be used. If the
data field bit count is not an integral multiple of eight,
additional zeros must be added to fill out the last 8-bit
quantity. The data field is followed by a 16-bit block
check for detection of errors in the data.
The block check is a 16-bit cyclic redundancy check
that uses the CRC-16 polynomial:
x(16)+x(15)+x(2)+ 1
It is the remainder after the division of the data stream
by the polynomial. All 48 bits of the header are included
in the header block check, and all bits in the data in the
data block check.
On the receiving side, the block check itself is included
in the computation, giving a zero result when no errors
are detected in the block being checked.
Protocol Messages
There are seven protocol messages plus two BOOT
messages for links not under standard DDCMP contro/.
The purpose of each message is described below.
ACKNOWLEDGE (ACK): Whenever acknowledgment is
required by a protocol message or a data message is not
forthcoming to acknowledge earlier data messages
received, an ACK is sent.
NEGATIVE ACKNOWLEDGMENT (NAK): If a data or
protocol message is rejected for any reason listed below,
a NAK is returned to the message transmitter from the
recl3iver. The reason for a NAK may be any of the
following:
Header block check error
Data block check error
REP message response
Buffer temporarily unavailable
Receiver overrun
Header format error
REPLY TO MESSAGE NUMBER (REP): A REP message
is sent from the transmitter to the receiver if the transmitter has not heard from the receiver within a predetE3rmined period and there are unacknowledged
messages outstanding. The response to a REP is either
an ACK (acknowledging the receipt of the last message
sent) or a NAK (acknowledging receipt of some preceding messages and indicating the need for
retransmission of succeeding ones.)

STaRT (STRT): Begins message exchange. The STRT
message is sent by one station to another to initiate
communications.
STart ACKnowledge (STACK). Transmitted by the
receiver to acknowledge receipt of a STRT message.
Off-Une Mode
DDCMP allows for a special off-line mode that is useful
for bootstrapping, dumping, and low-efficiency data
transfer for very small nodes (e.g., intelligent terminals
or small ROM link controllers). This mode ensures that
data will be checked for integrity via the CRC-16, but no
acknowledgement scheme is supported by the message
formats. Acknowledgement can be handled by a timeout.
mechanism or by a higher-level protocol enveloped in
the data portion of the message.
The message formats in this mode are similar to those
used in normal DDCMP messages.
Efficiency of DDCMP
The efficiency of DDCMP depends on the length of data
messages and traffic load on the link. Each DDCMP
data message requires ten characters for the message
header and block checks. Hence, the greater the
message size, the more efficient the protocol bcomes.
By including a response field in data messages, there is no
need for sending separate acknowledgement messages
in high traffic situations. This feature combined with the
ability to acknowledge up to 255 messages at a time
makes the protocol extremely efficient over long links
such as satellite links, and links with very high traffic
volume.
Other factors affecting efficiency on multipoint and
half-duplex configurations relate to the frequency of
selection and quantity of data transmitted per selection
interval. These can be traded off based on requirements
for throughput and responsiveness. In multipoint systems,
the control station does not have to complete error
recovery procedures with a given tributary before
selecting another one. This means that a single tributary
experiencing many errors will not particularly degrade
performance of the other stations on the link.
The protocol was designed to operate most efficiently
over links with some probability of introducing errors.
This includes virtually all links useful in DECNET. For
links which are extremely error free or noisy, this
protocol may not present the best solution to the
communication requirements. For these situations, such
techniques as forward error correction and assumed
acknowledgement with timeout may prove more efficient
for total data throughput on the link.
DDCMP supported modes are summarized in Table A2,
Appendix A.
3.4 THE NETWORK SERVICES PROTOCOL (NSP)
The logical link protocol layer is called the Network
Services Protocol. It is implemented in software by a
module called Network Control Services. Network Control
Services software routines must (1) allow general

communication between user-level tasks, and (2) do
general network supervision.

4. ECHO and ECHO RETURN-More maintenance
messages, this time aimed at the NSf' layer.
5. ROUTE-TABLE-Messages that allow dynamic
changes in the network structure (announce links
coming up or going down or other changes in network
structure, and name and describe nodes).

Inter-task communication breaks down into the following
subfunctions:
1. Create for the process layer a "virtual pipe" or
"logical link" through which to send and receive
messages. Logical links may extend through intermediate nodes.
2. Multiplex and demultiplex process-layer messages
on the logical links into physical links.
3. Control traffic through the logical link.
4. Ensure end-to-end delivery with correct sequencing.
5. Allow process interruption.
6. Destroy the logical link.

NSP supported features are summarized in Table A3,
Appendix A.
3.5 DATA ACCESS PROTOCOL (DAP]I

The Data Access Protocol (DAP) is a set of message
formats and rules used to transfer data between objects
in the network, namely, user processes, file storage
devices, card readers, line printers, othBr sequential
devices, and user-oriented terminals. Its main function
is to make access to remote devices and files convenient.

Network supervision breaks down into the following
subfunctions:

DAP operates under the Network Services Protocol.
Using NSP, a logical link is created between two objects
in the system. Once created, the link can be used by
the two objects to exchange messages for the purpose
of manipulation and transfer of data between the systems.
When data exchange is complete, the link is disconnected. The specific DAP functions include:

1. Inform the various network nodes of new nodes
coming up and give their characteristics,
2. Detect node or physical link failure and inform the
various network nodes of such failures or other
changes in network structure.
3. Route messages through the net.

1. Transfer of file data.
2. Transfer of accounting information.
3. File manipulation.
4. Transfer of file parameters.
5. Pass status over the link. examples:
File access complete
End of file
Illegal user identification
Illegal account information
Unsupported file attribute
Error on Device
Data error, transfer aborted

These are accomplished by defining for all messages a
general header that contains control and routing information. User-level data messages also contain a logical
address that corresponds to the logical link and a
message number that aids in sequencing and error
detection.
As in DDCMP, NSP also contains a number of control
messages that help request message transmission and
do general network supervision. These messages include
the following:
1. CONNECT and DISCONNECT-Messages that request creation and destruction of a logical link.
2. LINK STATUS and REQUEST LINK STATUSMBssages that allow request and acknowledgement
of data messages.
3. NO-OP (NOP)-A message that allows maintenance
work on the lower-layer DDCMP and interface drivers.

DAP is useful for transferring files to linls printers,
creating new files on remote devices, transferring files
from card readers or disks, and deletin!~ files and other
file structures.
DAP supported features are summarized in Table A4,
Appendix A.

DNA
"""'-

PROGRAM

DAP

NSP

---- -- - - -------D

NSP

DAP

PROGRAM

- ----- - ------

PDP-ll

PDP- '11

~----------------------------------------,,----------------------------------------~)
DEeNET
Figure 3-4
DNA Layers
16

CHAP-rER 4
SOFTWARE SYSTEM
COMPONENTS
and data management, and initiate the response to alter
and control on-going processes. There is no limit to the
number of tasks that can be stored within an RSX system.
Each such task is an independent entity that may be
installed, modified, or deleted independently of other
tasks.

This chapter describes the PDP-11 operating systems
that support DECnet: RT-11, RSX11-S, RSX-11 M,
RSX-11 0, RT11 and RSTS/E. These systems are all
part of the broad family of DIGITAL operating systems.
Each of these systems specializes in one or more areas.
RT·11 is a small, flexible system that will operate with
or without disk support in single-job or foreground/background mode. It is well-adapted to program
development and high-speed data acquisition. RSX-11 S,
'M and '0 make up a family of multi-programmed. realtime operating systems featuring great system power
and flexibility.

The basic program unit within RSX-11 is the task. Execution of a task may be initiated by the system operator
or by another task.
Once a task has been initiated, its execution and the
execution of all other currently active tasks are governed
by the RSX-11 Executive based on the occurrence of
internal or external events.

4.1 THE RSX-11 FAMILY
The RSX-11 family is a set of mUlti-programming realtime systems designed for a wide range of applications
where many processes must be monitored and controlled
concurrently. They provide easy coupling to on-line
processes through a variety of real-time interfaces, and
optimum real-time response through dynamic allocation
of system resources. They provide straight-forward
development of applications programs (tas'<s) through
on-line program development. And, they provide large
scale data management capability through an extensive
set of file processing techniques.

Input/Output
The RSX-11 I/O structure provides a flexible and efficient
mechanism for the support of standard PDP-11 peripherals and special purpose devices. This structure
provides input/output operations in parallel with task
execution, device independence, and easy integration
of new device handlers.

Event Flags and System Traps
The completion of an I/O transfer, or any other event of
significance to the system, may be signaled in either of
two basic ways: either by the setting of an "Event Flag"
or by a system trap. In the case of an "Event Flag," a
bit is set which can be tested by the running task or
waited for. A system trap interrupts (software interrupt)
the running task and forces execution of a separate subroutine.

The RSX-11 family includes three upward compatible
systems:
RSX-11 S -A small system designed to run stand-alone,
or as a satellite computer in a network.
RSX-11 M-A midsize system that includes all RSX-11 S
capabilities plus extensive file management
compatibility.

Data Management
In RSX-11 M and RSX-11 0, file handling facilities permit
the defining and accessing of named files, randomly or
sequentially, with functions such as space allocation and
directory maintenance handled automatically. These
functions are performed by the basic file control commands: OPEN, CLOSE, GET, PUT and DELETE.

RSX-11 D-A large-scale system that includes a batch
processor and multi-terminal operation.
Detailed descriptions are included at the end of this
section. Real-time operation is based on the interaction
of tasks stored within the system and events occurring
in the outside world. Examples of events include the
arrival of new data from an analog to digital converter,
a change in the status of a digital input point, the entry
of a. command from the system operator. Each event
constitutes a potential change in the status of the overall
system to which the RSX-11 system must respond.

Operator Control
The operation of the RSX-11 system may be monitored
and controlled by a system operator at any time. The
operator may (1) perform initialization functions (such as
entering the time and date), (2) request status information (such as a list of installed tasks), and (3) exercise
control over sysem operation (such as by starting and
stopping tasks).

Programs, or tasks, stored within the system provide the
response to real-world events. These tasks may be
written in either FORTRAN IV or FORTRAN IV PLUS
(supported by RSX-11 M and RSX-11 D), or COBOL (supported by RSX-11 0 only) or MACRO assembly language.
(RSX-11 S does not support program development; that
must be done under RSX-11 M and the task images then
shipped to the RSX-11 S system. They can monitor the
real-time input data, perform the necessary calculations

Distinctions among the RSX Family Members
RSX-11S
RSX-11 S is intended for real-time, multiprogramming
applications requiring an inexpensive computer with
minimal configuration requirements. The minimal con-

figuration includes 8K of memory; an extra 4K are
requimd of network support. Maximum memory size with
optional memory management support is 128K words.

DYNAMIC MEMORY ALLOCATION-Unlike the other
members of the RSX family, RSX-11 D can load a task
into any available memory. The task building (linkage)
process does not include memory assignment, because
that is determined at load time.

Although RSX-11 S can run in stand-alone mode, it is
ideally suited for a network environment, because the
netwotk can supply the file system and task-loading
capabilities that would not be possible in stand-alone
mode. Thus, RSX-11 S in a network could have all the
capabilities of RSX-11 M, except that file I/O and task
loading will usually be slower, simply because communications links are slower than internal data busses.

CORPORATE STANDARD BATCH-The RSX-11 D batch
facility provides the capability to submit requests for a
job or jobs to be performed in background mode. Each
job can consist of control information and data. The
Batch capabilities include:
1. Optimizing FORTRAN
2. ANSI'? COBOL
3. Shared Tasks
4. Dynamic Loading of Device Handlers

FORTliAN and MACRO programs can run under RSX11 S, but the system does not support program assembly
or cor1pilation. That, and system generation must be
done on an RSX-11 M or RSX-11 D system.

4.2 RT·11 FIB OPERATING SYSTEM

RSX·11M

RT-11 is a high performance Foreground/Background
operating system that combines powerful computing
hardware with user-oriented software. RT -11 is designed
for the single user involved in program development
and or real-time applications. It provides fast, simple,
on-line access to the full power of any DIGITAL PDP-11
processor with at least 16K words of memory plus mass
storage (24K words of memory for DECnet operation).

RSX-11 M is a medium-sized, disk-based operating
system that will run on any model of the PDP-11 with at
least ~~4K (10K without network support) words of
memory and one disk plus secondary storage that may
be disk, DECtape or cassette. As with RSX-11 S, memory
size can be as great as 128K words with memory
mananement.
In adolition to supporting all the features of RSX-11 S,
RSX-11 M supports 'full disk data management, checkpointing (roll-in/roll-out), and program assembly/
compilation.

RT-11's interactive nature and quick response offer
give-and-take flexibility plus rapid turnaround. Its queued,
real-time I/O and fast throughput take full advantage of
the outstanding real-time features of the PDP-11, such
as hardware stack processing, mUlti-level, mUlti-line
priority system, and vectored interrupts.

FILE STRUCTURE-The RSX-11 M file system is characterized by the following features:

The RT-11 F/B monitor allows two programs to operate
at the same time: a foreground program and a background program. The real-time function is accomplished
in the foreground which has priority on system resources.
Functions which do not have critical response time
requirements (e.g., program development) are accomplished in the background, which operates whenever
the foreground is not busy. Within their priorities, both
foreground and background are complete RT-11 systems
with access to all system functions. Although they operate
independently, foreground and background can communicate through disk files and/ or job communication
areas in memory.

1. On-line file allocation.
2. Comprehensive file protection.
3. Support of both random and sequential-access
mE!thods.
4. Support of both fixed and variable-length records.
Four levels of protection are available: System, User,
Group and Global. The file organization is independent
of the method of access; in other words, all files may be
accessed either sequentially or randomly.
CHECKPOINTING-RSX-11 M provides a checkpointing
option (roll-in/roll-out) whereby low-priority tasks can
be interrupted by high-priority tasks, copied onto the
disk to make memory available to the high-priority task
requesting execution, and later resumed from the point
of intHrruption.

If F/B operation is not required, the sin~lle-job monitorwhich requires less memory and overhead-can be
utilized. Should requirements change, upgrading is easily
accomplished since programs are completely interchangeable between the single-job monitor and the F /B
monitor.

PROGRAM DEVELOPMENT-RSX-11 M supports on-line
MACHO assembly and FORTRAN compilation, task
building and system generation. Load modules created
on an RSX-11 M system may be transmitted for execution
on RSX-11 S nodes.

4.3 RSTS/E TIMESHARING SYSTEM
RSTS/E is oriented to interactive rather than batch mode
operation. Users interact with the system through a
variety of terminals, and through resource-sharing, they
have access to and control over system peripherals such
as card readers, line printers, paper tape devices, disks
and magtape units.

RSX··110
RSX-l1 D is designed for efficient use of large-system
resources in an environment with rapidly-changing
needs. It supports everything RSX-11 M and RSX-11 S do,
and in addition, dynamic memory allocation, batch
processing, shared tasks, and dynamic device handler
loading.

Timesharing users interact with RSTS-E using BASICPLUS, a significant extension of Dartmouth BASIC. The

language is easy to learn and work with, yet puts the
power of the system at the users fingertips. The immediate mode of operation enables the terminal to be used
for simple calculations. Dynamic debugging is faster
since programs may be interrupted at any point, checked,
corrected, and operation resumed.
Normal timesharing use of RSTS-E consists of typing
program text using a keyboard terminal and at the end
of the program, typing a RUN command which initiates
program execution. A second mode (immediate mode)
consists of typing program statements on the keyboard
and having them executed immediately. Program statements are identical in either case except that, in immediate mode they are typed without line numbers.
Terminal user may have exclusive use of any peripheral
on the system (except the public disk(s) which is a
shared device). Not only can all devices be accessed
by users at any time, but any device can also be
accessed from a BASIC-PLUS program. For example,
one user may use the card reader, line printer, magtape
and disk for performing a "batch" administrative data
processing task; another terminal user may use a magtape unit for retrieving or creating a tape file intended
for off-line storage; and when the card reader is free,
yet another terminal user may read in a punched card
file which contains a BASIC program created at an
off-line card punch.

CHAF)TER 5
NETWORK BUILDING BLOCKS
5.1 NETWORK BUILDING BLOCKS

This maximizes host transaction processing throughput.

Some users may not require a full DECnet capabilityTheir requirements may only be for remote job entry
capability to an IBM system. Other users may wish to
move toward a full DECnet implementation in phases,
starting with relatively simple building blocks and
evolving toward DECnet gradually.

MULTI-TASK OPERATING SYSTEM (RSX-11 D) permits
a large number of different transaction application tasks
to be processed simultaneously.

DIGITAL offers a variety of packages which can be used
by themselves in communications-oriented applications
--or as building blocks toward a full network implementa.tion. The following network building blocks are
available:

EFFICIENT FILE SYSTEM provides a data base of over
700 million bytes. Overlapped seeks on multiple disk
drives minimize disk access time, the main bottleneck of
many transaction processing systems.

DECcomm600

MULTI-USER TASKS enable many terminals to simultaneously process the same applications task without
keeping multiple copies of the task in memory.

DYNAMIC MEMORY ALLOCATION and the ability to
checkpoint tasks on the disk while they are idle waiting
for another input permit maximum efficiency in use of
core memory.

a powerful multi-program, multiterminal system, which can serve as
the first step in a network application requiring support of multiple
terminals.

Remote Computer
Systems

RSX-11 M, RSX-11 D and RSTS/E
systems with facilities for transmitting
data to IBM host systems.

Front End
System Base

PDP-11-based system with hardware
and software for servi ng as a programmable front-end for IBM
systems.

TASK-TO-TASK COMMUNICATIONS capability enables
one transaction application task to activate a number of
other tasks if appropriate.
REMOTE CARD READERS and Line Printers are supported to enable entry of batch jobs.
DECcomm 600 provides an orderly expansion path from
simple to complex applications without requiring applications reprogramming. All applications written for a
single processor RSX-11 M/RSX-11 D system controlling
a small number of local terminals can be utilized unchanged on DECcomm 600. The DECcomm 600 package
provides turnkey communications software.

5.2 DECCOMM 600
DECcomm 600 is a multi-terminal system designed for
maximum effectiveness in transaction processing applications. It is designed to permit a large variety of users,
who may be geographically dispersed, simUltaneous
access to a computer data base. Each user has his own
interactive terminal by which he may enter new items
into the data-base, inquire of existing information or
modify and update the data base.

When the applications requirements have expanded to
where remote data bases, rather than remote terminal
clusters, are needed the applications code can be moved
out to these sites from the DECcomm 600 host. DECnet,
DIGITAL's computer network facility provides the capability for these remote data base sites to communicate with
the DECcomm 600 host.

DECcomm 600 contains RSX-11 D plus communications
extensions to permit multiple terminal support and
clustering of terminals at remote sites; and it can be
used to extend DECnet to support of terminals.

Host system

The host system can consist of any valid RSX-11 D
system (PDP-11/40, PDP-11/45, or PDP-11/70). The
PDP-11 /40 should appeal to economy minded users,
while the advanced features of the PDP-11 /70 should be
particularly appealing to users for whom high performance is important. Any peripherals supported by
RSX-11 D can be used; but of particular significance are
RP04 disks which can support substantial data bases
(88 million bytes per drive).
DECcomm 600 uses version 6A of RSX-11 D.

The special features of DECcomm 600 which make it
highly effective in this environment are:
TURNKEY COMMUNICATIONS PACKAGE provides
transparent support between the host and both local
and remote terminals. Both operator interaction and
applications code are unchanged regardless of terminal
location.
COMMUNICATIONS COST SAVINGS are provided for
clusters of remote terminals through the use of concentrators.

Terminal Concentrator

REDUCED COMMUNICATIONS OVERHEAD in the Hostconcentrators communicate with the host on a Direct
Memory Access line-by-Iine basis. All character-bycharacter terminal handling is done by the concentrators.

PDP-11 /10's or PDP-11 /40's functioning as data concentrators, are used to control communications between
the host computer and remote terminals. The data concentrators help reduce line costs by accepting messages
21

from many terminals and transmitting them to the host
processor via a single high-speed line.

The clustering of terminals at remote sites to reduce
communication costs

subject to noise interference. DDCMP is designed to
operate over full- and half-duplex channels in point-topoint and multipoint modes, independent of the bit width
and other characteristics of the data channel. It is applicable to multiple computer configurations such as
computer networks, host/front-end processors, remote
concentrators, and remote entry/exit systems.

Message assembly and disassembly and character
interrupt handling to free up the RSX-11 D system

Total Terminals Per Host:

64 Standard; more with
additional. software

A PDP-11 concentrator running under TC/D relieves the
RSX-11 D host of the processor load inherent in servicing
large numbers of character interrupts. Under TC/D the
concentrator accepts and stores characters received from
terminals until it receives a control character (e.g.,
carriage return), indicating the end of a complete record.
It then transmits the complete record over a synchronous
communications link to the host processor at speeds up
to 9,600 bits per second.

Total Concentrators Per Host:

Total Card Readers:

1 per concentrator

Host-Concentrator
Transmission Speeds:

9600 bps maximum
When 4 Te/D's are used
ma;:imurn aggregate
speed is 19.2K bps with
no single line to exceed
9600 bps

The communications architecture is simple, point-topoint transmission, making it easy to design a network
to suit lindividual requirements. Transmission between the
remote terminals and host computer is full-duplex, enabling simultaneous transmission in both directions.
Terminals may be connected directly to the host processor, or over switched or leased common carrier
facilities using modems.

Upgrading OECcomm 600

The concentrator allows:
Sirnultaneous use of many interactive terminals

Performance Summary

The use of remote line printers and card readers

A DECcomm 600 system can serve as the initial step
toward a DECnet implementation.
Upgrades to DECcomm 600 could include:

A. Upgrading the remote concentrators (PDP-11 /1 O's
or PDP-11 /40's) to contain RSX-11 M software. Communications between the host RSX-11 D and the
satellite RSX-11 M would be via DECnet facilities. The
satellite RSX-11 M could perform local processing
and/or support a local data base.

Half-duplex transmission (I.e., transmission in either
direction but not in both directions simultaneously) is
provided as an option. This option permits the use of
dial-up connections between the host computer and
concentrators and between the concentrators and
terminals.

B. Connecting multiple DECcomm 600 systems at
different sites using DECnet facilities.
5.3 REMOTE COMPUTER SYSTEMS

An automatic dialing feature enables the host computer
to make its own telephone calls to terminals. Any
telephone number known to the applications program
can be dialed by the user simply by issuing an I/O
statement.

Remote Computer Systems are versatile general-purpose
packages which can be used in a wide variety of ways.
Remote computer systems have been used for applications requiring powerful local processing and transmission of files to a host IBM system, for applications
requiring transmission of magnetic tape data, and for
applications requiring local data base facilities.

Complex calculations for terminal polling sequences, line
10adin~I, and queuing theories are not required; each
terminal performs as if it has a direct private line all the
way to the host computer.
Concentrator-to-terminal communication is simple asynchronous transmission at speeds up to 9,600 bits per
second. Host-to-concentrator communications uses
synchronous transmission techniques at speeds up to
9,600 bits per second.
Direct Memory Access (DMA) 1/0 interfaces are used
on the host PDP-11 IRSX-11 D side of the communications link to handle the character servicing requirements
and optimize the transaction processing capability of the
host system.
Transmission between the host PDP-11 IRSX-11 D processor and the remote PDP-11 concentrator(s) makes
use of DIGITAL's new message-oriented protocol
DDCMP (DIGITAL Data Communications Message Protocol). DDCMP assures the correct sequencing and data
integrity of messages transmitted over data channels

RSX11 12780

RSX-11 M or RSX-11 D real time operating
systems with 2780 emulation. Transmission can be to and from PDP-11 disk and
tape mass storage. 27801 communications
can be concurrent with other RSX-11
processing.

RSTS/2780

PDP-11 timesharing system with 2780
emulation. Transmission can be to and
from PDP-11 disk and tape mass storage.
2780 communications can be concurrent
with other RSTS operations. RSTS terminal users can name fi\l:~s to be transmitted.

A Remote Computer System can:
Transmit to and receive from mass storage for
greater efficiency
Permit use of unit record equipment simultaneously

with data transmission (RSX11 0/2780 and
RSTS/2780)

RSX-11 0/2780

RSX-11D is the general-purpose multi-tacking operating
system for the larger members of the PDP-11 family. It
offers task scheduling, task protection, file-structured
I/O, program preparation, and utility features normally
associated only with much larger computer systems.

Be an on-the-spot data processing center
Reduce communications costs by processing some
data locally and transmitting only essential input
to the IBM computer

RSX-11D is applied to a broad spectrum of scientific
computer, laboratory process control, and transactionprocessing environments. RSX-11 0/2780, running as a
task in an RSX-11 0 system, extends the range of applications of RSX-11 0 into situations where on-line interface to an IBM computer is required.

Support its own data base for quick access
Be upgraded to include DECnet capabilities
The Remote Computer Systems will communicate with
an IBM system containing HASP, ASP, POWER, or RJE
programs. The IBM system should leave a 2701 data
adaptor, a 2703 Transmission Control Unit, a 3704 or
3~ro5 Transmission Controller or a System 370 Model 135
Integrated Communications Adaptor.

Input to RSX-11 0 can be processed or reduced, and
organized into files, and then transmitted to an IBM
computer with RSX-11 D/2780. The system operator
controls RSX-11 0/2780 operations via console commands. Input can be transmitted from an optional card
reader or from as many as six RSX-11 D files by a single
console command. Received data can be printed directly
on a line printer or routed to RSX-11 0 mass-storage files
for later manipulation and printing. RSX-11 D-based
Remote Computer.

The following 2780 features are emulated:
400-byte buffer (maximum transmit or receive block
size)
Multiple record feature (up to seven logical records
per block can be transmitted or received)
ASCII or EBCDIC-transparency code (console
selectable)

5.4 IBM FRONT END SYSTEM BASE

Automatic answer

The DECcomm Front-End System Base (FSB) consists
of a PDP-11 family computer, an IBM 360/370 interface,
and DIGITAL communications software, and is intended
as the major element in a 360/370 programmable frontend processing system. It is designed to increase system
capability for less cost, to increase application flexibility,
and to interface terminals not supported by standard
IBM equipment.

Horizontal printer format control
Extended retry on data-link errors
RSTS/E 2780
RSTS/E 2780 combines the support for 31 interactive
terminals with the ability to manage data bases upwards
of 300 million-characters and the ability to transmit files
to a host 360 or 370.

With the DECcomm FSB, the user can build a front-end
communications system fully compatible with OS/GAM,
OS/TCAM, DOS/QTAM, and DOS/BTAM. With the
attention interrupt capability, polling operations can be
removed from the 360/370 set of tasks. The front-end
system interrupts the S/360 when it has data to present.

With an interactive terminal, each RSTS/E user can
request named mass storage to be transmitted over the
2-'80 link. RSTS/2780 queues all such requests and
sHrvices them on a first-in, first-out basis. RSTS/2780
places each file received over the link on mass storage
under a separate file name. Users can then interrogate
RSTS/2780 at any time to determine the names of files
queued for transmission.

When designed with modular COMTEX program units,
the FSB facilitates the construction of small to large
communications systems. A 2848 Emulator Terminal
Application Program (ETAP) is provided to make the
hardware look like a 2848 display controller to the S/360
or S/370. The compatibility is such that standard IBM
diagnostics will run, including the 2848 On-Line Test
Program (OLTEP).

In addition to the user-oriented "queued" mode, RSTS2-;'80 can be controlled interactively by the RSTS/E
system operator. By typing commands at the console,
the operator can indicate files to be transmitted immediately and can dynamically specify the destination of
received files.

Front Ends and Networks

With RSTS/E, every terminal user can access all system
peripherals and resources. Even when RSTS/2780
operations are in progress, RSTS devices and resources
continue to be shared by interactive users.

The Front End System base can be used as an element
in a network. However, it is important to note that software (COMTEX) currently supplied with the FSB package
is not supported under DECnet. The FSB requires user
applications software and "customization" for use in
networks.

RSTS/E systems can be built around PDP-11 /40 or
PDP-11 /45 computers with up to 248K bytes of memory
and disk storage of up to 343-million bytes. Users interact
with the system through BASIC-PLUS, an easily learned
interactive extension of Dartmouth BASIC language.

PART II
COMMUNICATIONS

HARDWARE
This section provides information to assist in selecting
PDP-11 :

communications interfaces
modems
It is important to note that all the products described in
this section are not necessarily supported by DEenet.

CHAP-rER 6
HARDWARE COMPONENTS
6.1 PROCESSORS

PDP-11 Processor Handbooks. The following table provides some guidelines in differentiating PDP-11
processors:

Any PDP-11 family processors can be used in DECnet
applications. The reader is advised to consult other
documents in selecting processors-particularly the

Table 6-1
Summary of PDP-11 Processors

PROCESSOR COMPARISON TABLE
MINICOMPUTERS

SYSTEM COMPUTERS

PDP-11/04

PDP-11/05

PDP-11/10

PDP-11I35

PDP-11/40

PDP-11/45

PDP-11/70

MAIN MARKET

OEM

End User

OEM

End User

End User
and OEM

MICROPROGRAMMED
ST ACK PROCESSING
PROGRAMMABLE
STACK LIMIT

yes
yes

opt

yes

GENERAL REGISTERS
REG-TO-REG TRANSFER
HARDWARE FLOATING
POINT

8
2600 nsec

8
2700 nsec

8
900 nsec

16
300 nsec

32-bit
(opt)

32,64bit (opt)

MAX MEMORY SIZE
(BYTES)
MEMORY TYPE

56K
MOS

56K
CORE

248K
CORE

248K
BIPOLAR
MOS
CORE

2M
BIPOLAR
(cache)
CORE
(main)

EFFECTIVE MEMORY
SPEED

500 nsec

980 nsec

1000 nsec

400 nsec

MEMORY PARITY
MEMORY MANAGEMENT
PROCESSING MODES

no
no
1

opt
opt
2 (opt)

yes
opt
2 (opt)

300 nsec
500 nsec
1000 nsec
yes
yes
3

yes

REAL-TIME CLOCK
PROGRAMMER'S CONSOLE
HARDWARE BOOTSTRAP
SERIAL LINE
CONTROLLER

yes
opt
yes

yes
yes
opt

opt
yes
opt

yes
yes
yes

yes

WARRANTY

30-day
return to
factory
opt

gO-day
on site

30-day
return to
factory
opt

90-day
on site

INSTALLATION, ON-SITE

30-day
return to
factory
opt

yes

DIAGNOSTICS
MAINTENANCE
MANUALS

with first
system

yes

with first
system

yes

AUTO HARDWARE
INTERRUPTS
AUTO SOFTWARE
INTERRUPTS
POWER FAIL/AUTO
RESTART

yes

NO = NOT AVAILABLE
YES = IS STANDARD AND IS INCLUDED
OPT = IS OPTIONALLY AVAILABLE

yes
yes
3

6.2 ASYNCHRONOUS LINKS

• set for odd parity
• set for even parity

An asynchronous interface is used to connect a computer directly to a terminal or to a remote terminal via a
modem and a telephone line. In some applications
PDP-1'I's can be connected using asynchronous links.

Depending on the interface, the parity
bit is selected by jumpl3rs, by switches
or by the program.

Asynchronous transmission normally involves the
transmission of characters (from a keyboard and to a
terminal's printer or CRT) at random intervals. Each
character is separated by STOP AND START bits.

FULL HALF
DUPLEX:

In selecting an asynchronous interface, the following
must be taken into consideration:
Type of connection required (direct terminal connection, private phone line, DDD phone line)
Ability of interface to perform such things as parity
checking, insertion and stripping of start/stop bits,
buffering, reducing per character interrupts

MODEM
CONTROL:

Ability of software to modify the interface so as to
change line speeds, etc.
Typical features which mayor may not be included in an
asynchronous interface:
DATA 131TS:

STOP BITS:

BAUD RATE:

PARITY BIT:

Some asynchronous interfaces can have
a variable number of data bits (5, 6, 7,
8, etc.) in order to handle different
communications codes (such as ASCII,
Baudot, EBCDIC). There are three ways
the code$ can be set depending on the
interface; by inserting or removing
jumpers (wires), by manually setting
switches in the interface hardware, by
software.

INTERRUPTS:

Generally one or two stop bits are used.
Slow speed transmissions (up to 100
baud) normally use 2 stop bits; faster
transmissions (150 baud and higher)
normally use one stop bit. Some IBM
type terminals (2741's) use one and one
half stop bits. Depending on the interface, stop bits can be jumper, switch
or program selected.

Full-duplex transmission implies simultaneous bidirectional communications;
half duplex implies transmission in one
direction at a time. Depending on the
application and the type of line used
half- or full"duplex transmission is used.
Some interfaces include provisions for
either type of transmission.
Two basic types of modems are available: For connections to private telephone lines, and for connections to the
standard dial-up telephone network.
When the dial-up telephone network is
used a variety of signalls must be exchanged between modems (ring, carrier
ready, etc.). Some interfaces can
control DOD modem signals, while
others are only capabl'9 of being connected to private line modems.
Normally each time a character is
received or transmitted an interrupt
must be issued by the program. Some
interfaces have provisions to place or
remove blocks of characters in or out of
memory without issuinlg an interrupt
for each character.

6.3 PDP-11 ASYNCHRONOUS INTERF:ACES

The following PDP-11 asynchronous intl3rfaces are
available:
DL 11 SINGLE ASYNCHRONOUS LINE IINTERFACE
The DL 11 is used to connect a PDP-11 to a terminal or a
single line. A variety of characteristics can be selected
(no. of stop bits, parity, full- or half-duplex, data bits,) but
cannot be changed under software control.

Asynchronous interfaces contain a
crystal clock which determines the
transmission rate. Depending on the
type of interface, the baud rate may
either be fixed at a single speed when
the interface is installed or changed by
the program.

DJ11 16-L1NE ASYNCHRONOUS INTERFACE
The DJ11 is used to connect terminals or lines to a
PDP-11 in 16 line groups. A wide variety of characteristics can be preselected but cannot be modified under
software control. It contains a 64-character "silo" buffer
which can be used to reduce interrupt overhead for
incoming characters.

Some applications require some form of
checking in order to determine whether
the character has been received correctly. In these applications a parity
check is performed through the use of
a parity bit which is added to each
character. Depending on the terminal
the parity bit can be:

DH11 16-L1NE ASYNCHRONOUS INTEF~FACE
The DH11 is used to connect terminals or lines to a
PDP-11 in 16-line groups. A wide variety of characteristics can be selected under program control, so
terminals with different characteristics and transmission
speeds can access the PDP-11. It contains a 64character buffer for incoming characters, and has DMA
capability for transmitting characters.

• not used

Feature Comparisons
DL11

DJ11
16

DATA TRANSFERS:

CHARACTER
INTERRUPT

SILO (Input)

SILO (Input)
NPR (Output)

SPEEDS:

Up to 9600 bps
(Fixed)

Up to 9600
(Fixed in
4-line groups)

Up to 9600
Program-Selectable
On Each Line

FULL OR HALF
DUPLEX:

YES

RECEIVE BUFFER:

DOUBLE-CHARACTER

64-Character
SILO

SPLIT SPEEDS:

YES

CHARACTER SIZES:

5, 6, 7, 8, bits
STRAP-SELECTABLE

5, 6, 7, 8, bits
SWITCH-SELECT

5,6,7,8, bits
PROGRAM-SELECT

PARITY:

ODD, EVEN, NONE
STRAP

ODD, EVEN, NONE
SWITCH

ODD, EVEN, NONE
PROGRAM

STOP BITS:

1, 1.5, 2
STRAP

1, 1.5, 2
SWITCH

1, 1.5,2
PROGRAM

20 MAMP

DL 11-A, DL 11-C

DJ11-AC

DH11-AA + DM11-DA

EIA (LEASED)

DL 11-B, DL 11-C

DJ11-AA

DH11-AA + DM11-DB
or DH11-AD

EIA (DDD)

DL 11-E

DH11-AE or DH11-AA
plus DM11-BB plus
DM11-DC

YES
DH11-AA + DM11's

Number of ·Lines
Per Interface:

MIXED 20 MAMP lElA

DH11

CABLES INCLUDED

YES

DH11-AA + DM11: YES
DM11-AE, AD: NO

AUTO ANSWER

YES

AUTO DIAL OPTION

+ DN11

+DN11

6.4 CONFIGURING SYSTEMS WITH ASYNCHRONOUS
INTERFACES

to 20mAmp and EIA terminals or to leased or DOD line
modems.

DL,11

It consists of a single SPC board and a cable. It will
mount in any PDP-11 standard SPC slot.

The DL 11 can be used to connect the PDP-11 directly

l1-B~ EIAICCITT

OL
OL 11-0
OL l1-E

DLll-A} 20M AMP
Dlll-(

:29

Special Considerations:

No more than 32 DL 11's
should be mounted in the
same PDP-11 cabi net.

Cables:

Cables are provided. In some
instances cables can be
extended to greater distances
-See DECcomm Hardware
Guide for DF11's.

terminals and modems using the same 16-line DH11.
DM11 modules must be added.
The DH11-AE permits connection to 16 EIA terminals or
private line modems.
The DH11-AD permits connection to 16 EIA terminals or
DOD line modems.
DH11-AA

Consists of a double system unit, which mounts in the
PDP-11 CPU or mounting box, and a DM11 type
distribution panel. DM11 modules must be added to
the distribution panel for line connections.

DJ11

The DJ11 can be used to connect the PDP-11 directly
to 20mAmp or EIA terminals or to leased line modems.
It consists of a system unit and a distribution panel. The
system unit will mount in a PDP-11 CPU and the
distribution panel requires 5 112 inches of rack space
(SM PAN).

Cables:

DH11-AA
DOUBLE SU

OJ 11 SU

MALE ElA/CClTT
CONNECTORS OR
4-SCREW TERMINAL
STRIPS
CABLES:

DM11 modules include cables.

DM11 MODULE SET

No cables are provided with the DJ11. For
20mAmp connections the distribution panel
has 16 4-screw terminal strips. Cables
from terminals can be wired directly to the
distribution panel screws.
For EIA/CCITT connections BC05-D-25
cables can be added separately.

DH11

The DH11 can be used to connect the PDP-11 directly
to 20mAmp and EIA/CCITT terminals and leased line or
DOD Ii ne modems.

DISTRIBUTION PANEL
FOR DM 11 MODULES

The DH11-AA permits mixed connections to 20mAmp

extensive checking techf1iques are used to recover from
possible transmission-in~uced errors. Synchronous
equipment is more expe~sive but is often used where
data integrity or performJnce is more important than
ultimate low cost.
'

DHI11-AD, DH11·AE

Consists of a double system unit which mounts in the
POP-11 CPU or mounting box and OJ11-type distribution
panel which requires 51/4 inches of rack space.
CABLES:

OH1'I-AO and OH11-AE do not include
cables (Order BC050/25).

Selecting a Synchronous Interface

The selection of a synchr'anous interface depends on
various factors:
• transmission speed raquirements
• interrupt loading on tHe processor
• ability to relieve proce~sor load by inserting and
removing control characters, performing BCC, etc.
• cost
• mounting space required
6.6 PDP-11 SYNCHRONOUS INTERFACES

Three synchronous interf~ces are available for use with
the POP-11, the OU11, th;e 0011 and the OV11. These
are briefly described in t~e following paragraphs. Complete descriptions are contained in the respective option
bulletins.
;
SINGLE LINE
CHARACTER
INTERRUPT

OU11!

OMA

0011

MULTIPLE LINE

OV11

DU11 Synchronous Inte~face

The OU11 is a low-cost synchronous interface. It
consists of a single small! peripheral control board
(SPC) and consequently ~equires very little mounting
space. It does not perforrr NPR data transfers. For
CRC/LRC checking, the KG11 is used in conjunction
with the OU11 (or software CRC calculation can be done
where CPU overhead is n:ot an important consideration).

DJ 11- TYPE DISTRIBUTION
PANEL. REQUIRES BCOD-25
CABLES.

DQ11 Synchronous Inte~face

The 0011 is a high-perfdrmance NPR synchronous
interface. It also has opti¢ms for CRC/LRC checking
(0011-AB) and special protocol character handling
(0011-BB). It is capable bf handling very high speed
transmissions (up to 1 megabaud, depending on
message length).

6.5 SYNCHRONOUS LINKS

Synchronous transmission is the transmission of coded
information according to a predetermined synchronization-bits always appear at the same time intervals.
When the line is open and there is no data to transmit,
some form of "idling" is maintained in order to retain
synchronization (or you resynch following the idle
period). Synchronous transmission is contrasted with
asynchronous transmission in which a continuous stream
of bits does not have to be maintained. In asynchronous
transmission each character is delimited by start or stop
bits, and no data is sent between characters.
Synchronous transmissions are generally used when
blocks of data messages, streams of characters, etc.
are to be transmitted. Computer-to-computer, remotebatch terminal-to-computer, intelligent terminal-tocomputer-usually use synchronous communications.

DV11 Synchronous Preprocessor
;

The OV11 permits conne9tion of 8 or 16 synchronous
lines to a POP-11. It cont$ins a microprocessor and
relieves the POP-11 of special character handling and
BCC generation. It is cap!able of NPR transfers on all
lines on both transmission and reception.

Synchronous transmission is generally more efficient
than asynchronous transmission because less bits are
used (no start/stop bits for each character), and more
:'31

Featur4! Comparison
OU11

0011

OV11

CRC/LRC
Error Checking

KG11 (Unibus
option--1 KG11
can service multiple
OU11's)

0011-ABInternal option
for 0011; no load
on POP-11. One 0011 -AB
required for each 0011.
(KG11 can also be used)

Internal on all
OV11's. All
CRC/LRC handled
within OV11. No
load on POP-11

Speeds

up to 9600 bps

up to 1 Megabaud

38,000 cps total
throughput for all 16 lines
(each line up to 9600 bps
full duplex)

Mounting

1SPC

0011-0A
1 SU
0011-AB
1 SU
0011-BB mounts in A8

OV11-AA
OV11-BA

NPR

Yes

Yes All lines

=
=

Total Lines
Per Interface

= 2 SU
= SM PAN

Full or Half
Duplex

Yes

Character -Sizes

5, 6, 7 or 8 bits

up to 16 bits

8 bits

Sync Character
Stripping or
Insertion

Programmable

Auto Answering

Yes

Parity-checking
and generation

Yes
in hardware

Programmable
Modem Control

Yes

OFC11
option

0011-KA
option

Internal for
all lines

Sync Character
Selection

Programmable

Programmablle
selection of 11 of
2/4-line group

Special Protocol
Character Recognition Hardware

0011-88
option

Internal for
all lines

8ell 303
Interface

OF11-G

0011-EA

Clock

8us Loads

OU11-0A
KG11
1

0011-0A
D011-A8

=1
=1

OV11-AA
DV11-A8

= ~~
= ()

6.7 CONFIGURING SYSTEMS WITH SYNCHRONOUS
INTERFACES

• maximum aggregate data rate of data being
received and/or transmitted

Throu!~hput

• the number of interrupts generated by the
reception or transmission of data

One 01' the first considerations in configuring systems
with synchronous interfaces is to determine whether the
POP-11 system can handle the data rates of the
communications lines. Total throughput is the total
data, ~Ienerally calculated in characters per second,
which can be handled by the system. A throughput
calculation must take into account:

• the amount of CPU intervention required
• length of message being transmitted
• CPU availability
• effect of peripherals
• capability of the operating system
32

Since throughput calculations are dependent on many
variables they vary tremendously from application to
application.
At one end of the scale, the capabilities of the PDP-11
operating system must be considered.
At ttle other end of the scale, one might encounter
applications where the user will develop his own
software system and dedicate the system to handling
communications. In these situations throughput
capabilities will possibly be substantially higher.

DQll-AB

Mounting Requirements

CRC/LRC OPTION

Standard PDP-11 configuration rules apply to the
DU11, 0011 and DV11 as follows:

-':~-=

DU111

DQ11-B8:
SPECIAL
CHARACl1ER
OPT.

The DU11 can be used to connect the PDP-11 directly
to a single leased or DOD line modem.
It consists of a single SPC board and a cable, and will
mount in any standard PDP-11 SPC slot.

-Jr~~:~~
DQ11-K

DUll-DA

CL~CK OPT.

__ rL,__ JL___J~_.

Special Considerations:

Cab!les:

The KG11 is available for use
with the DU11 for hardware
CRL/CRC calculations.

Special Considerations:

INote options (D011-AB, BB,
iKA) illustrated above. 0011's
ican be mounted in any PDP-11
ICPU or mounting box except
fDP-11 /20 and BA11-ES).

Cables:

;The D011-DA includes a 7.6m
:(25-feet)
cable.
;

DV11

;

The DV11 can be used to ~onnect private or DOD line
modems to a PDP-11 in 8 ¢r 16 line groups. It consists
of a double system unit (OY11-AA) and a distribution
panel (OV11-BA).
'

The DU11-DA includes a 7.6m
(25-feet) cable.

DQ11
The 0011 is a high performance synchronous interface
for connecting a private DOD line modem to the PDP-11.
Several 0011 's can be used as a single PDP-11.

;:;;;;;;...~D~Q~l~l·~-~DA SYSTEM UNIT

DV11-AA MAIN UNIT
2SU
33

6.9 PARALLEL LINKS
PDP-11 's can be connected (UNIBUS-TO-UNIBUS)
locally through the use of a DA 11 parallel interface.
The DA 11-B is a DMA link for high speed parallel data
transfers between two PDP-11 's using the directmemory-access (DMA) facilities of each computer, the
link transfers either single words or blocks of data from
the memory of one machine to the memory of the other.
Transfer rates can be as high as 500,000 words per
second.

DVll-BA DISTRIBUTION PANEL
Special Considerations:

Cables:

6.10 DISTANCE CONSIDERATIONS

DV11 AA can be mounted in
an 11/40, 11/45 or 11/70 CPU
or H960 series mounting box.

Parallel Connections (using DA 11-8)
The maximum distance is 30 meters (100 feet).

Must be ordered separately
(BC05-D-75) .

Asynchronous Connections
20mAmp: Distances can vary considerably depending
on the type of terminal used, the type of computer
interface and the transmission speeds used.

Modem Connections
Private Lines
Modem

DOD Lines
Modem

DQ11-DA, DU11-DA
DV11-AA (+DV11-BA)

Model (Data Rate)
201 A (2000 bps)
201 B (2400 bps)
201 C (2400 bps)

Model (Data Rate)
201A (2000 bps)

208B (4800 bps)

DQ11-EA

208A (4800 bps)
209A (9600 bps)
303 series
(19,200 bps to
50,000 bps)

The following table shows transmission speeds possible
at various distances when two PDP-11 's are connected
using DF11-K adapters.

201 C (2400 bps)

110,300 BAUDS AT 3000 FT.

SAME POINT
(1000Fn

6.8 SYNCHRONOUS LINK EXAMPLES
1. Connection via Modems

SAME

DUll
DQll
OR
DV 11

SAME
/POINT

DUll
DQll
OR
DV 11

~--------------~

2. Local Connection without Modems

0.3

1.2

2.4

4.8
DATA RATE, K BAUDS

9.6

o Cable is 3 twisted pair, no. 22 AWG stranded, each
~

~------------~
MAX. DISTANCE =50 FEET (AT 9600 BAUD)

pair shielded, Belden no. 8777, DEC PIN 91-07723;
shields connected to common leads at both ends.

3. Local Connection without Modems (high Speed)
(0011 )

o Cable is 2 pairs, no. 22 AWG solid inside station wire,
DEC PIN 91-05856-04 (Ouad).

Cable is Bleden no. 8777 as above, except shields are
Not connected at either end.

4. Local Connection without Modems
(DV11 /DU11 /0011)

EJ--1

DQll-KA

H312 •

Note: Circuit is DF11-K to DF11-K with transmitters
active and receivers passive. Distortion limit is 10 percent.

EIA/CCITT: The standard rule when modems are used
is that the maximum distance (between the computer
and the modem or between the terminal and the modem)
cannot exceed 15 meters (50 feet).

IOVll ••

IOVll*J--1

H312·

OQll-KA

t----8

However in some instances EIA/CCITT asynchronous
interfaces can be used to connect two PDP-11 's

* Re·quires wiring

modifications
* * Contains internal clock

together without modems. The following table provides
some distance guidelines.

Synchronous ConnectiOns

With Modems:

Whenever modems are used the
EIA/CCITT 15 meter (50 foot) rule
applies.

Without Modems:

When two PDP-11 's are connected
using synchronous interfaces without
modems the maximum distance
between PDP-11's must not exceed
15 meters (50 feet).

:r:

1-.

6.11 SPECIAL ADAPTE~S

lJ.1

~:
~

1
1.)0.5

A variety of adapters are available for use with the
PDP-11 synchronous and asynchronous interfaces.
These include:

SAME
POINT

__
__ +I_ _ _ _ _ _
________
1.2
2.4
4.8
DATA RATE. K BAUDS
~

_ _ _ _ _ _+I_ _

DF11-G:

Permits connection of DU11-to-Bell 303
series interfaces.

DF11-A:

Permits EIA/CCITT connections between
PDP-11 'so

DF11-F:

Provides TTL-to-20mAmp conversion.

DF11-K:

Provides TTL-to-20mAmp conversion and
permits asynchronous communications
between PDP-11's.

H312A:

Notes: Signal tests were between DF11-A's, with distortion limit of 10 percent. Max ground differential is 2 V
between ends.

Permits connection of two male type
EIA/CCITT connectors (used for local
connections between PDP-11 's without
modems).

DFC11-A:

Provides clocking when two PDP-11 's are
connected locally in synchronous mode
without modems.

These figures are also applicable to DL 11-B, 0, and E;
DJ11-AA.

H319:

Permits connection locally between two
PDP-11's in asynchronous mode.

0.3

9.6

DF11-A active element is M594 module.
6

Cable is 2 pair no. 22 AWG solid inside station wire.
DEC PIN 91-5856-04 (Quad).

o Cable is 3 twisted pair, no. 22 AWG stranded, each
pair shielded. Belden no. 8777, DEC PIN 91-07723;
shields connected to common (ground).

6.12 DECNET INTERCO:NNECTIONS

*Tl1is taple is for a voltage interface for private installations
not requiring conformance with EIA standards.

Table A1, Appendix A, summarizes maximum DECnet
transmission speeds for ~ach DEC operating system, and
Table A2 summarizes int~rconnectability capabilities
(asynchronous, synchronous, parallel) between
operating systems.

CHAPTER 7
MODEM SELE:-CTION GUIDE
SELECTING MODEMS

warranted to work with the equivalents of Bell modems.
However, specific modem equipment depends on the
country and the type of equipment in use in that country.

In North America, Digital Equipment Corporation tests
and warrants communications products with Bell
modems. Generally DIGITAL documentation specifies
appropriate Bell modems or "their equivalents."

The following Bell modem tables are provided for
guideline purposes.

Outside North America, DIGITAL products are generally

7.1 ASYNCHRONOUS MODEMS
TYPE MOIPEM
RECOMMENDED
SPEEDS

PDP-11 INTERFACE

PRIVATE LINE

DDD

Up to 300 Baud

DL11, DC11, DJ11, DH11

103A, E, F, G
113A, 113B

103A, E, R

300 to 1800 Baud

DL11, DC11, DJ11, DH11

202C, D, E, R, T

202C, D, E, S

7.2 SYNCHRONOUS MODEMS
2000 Bps

DU11, D011, DV11

201A, C

2400 Bps

DU11, D011, DV11

201C

201 B, C

4800 Bps

DU11, D011, DV11

208B

208A

9600 Bps

DU11, D011, DV11

209B*

7.3 SYNCHRONOUS (WIDEBAND) MODEMS

* Permits several lines to be multiplexed. Combinations are:
1

19,200 Bps

DU11 /DF11 G, D011-EA

303B

50,000 Bps

D011-EA

303C

1 9600 bps line, 24 800 bps lines, 4 2400 bps lines.

7.4 MODEM OPTIONS
Bell modems offer a variety of optional features
depending on the type of modem and specific modem'
model. The following table indicates some optional
features and DEC recommended choice.

Con of Carrier
Disconnect Yes/No

Yes

Space Disconnect
Yes/No

Yes

EIA/Contact

EIA

Automatic Answer
Yes/No

Yes

New Sync
With/Without

Without

Carrier Control
Continuous/Controlled
by RTS

Continuous

Alternate Voice

Depends on customer
requirements

OPTIONAL FEATURE

DEC RECOMMENDED

With or Without
Auto Calling Unit

Depends on application

Perm. Wired or Key
Controlled Auto Answer

Key Controlled

Terminal Responds
To Disconnect (Yes/No)

Yes

Terminal Initiates
Disconnect (Yes/No)

Yes

Mark Hold/Space Hold

Mark Hold

Answer Mode Indicator
Off/On

Off

Clock Source
Internal / E>--ternal

Internal

Answer Control
Separate/Combined

Combined

Carrier Control
Conti nuous / Switched

Switched

Send / Disconnect
Yes/No

Yes

Request to Send Control
Conti nuous/ Switched

Switched

APPENDIX A

DECNET INTERCONNECTION FACILITIES

MAXIMUM
ASYNCHRONOUS
RATE

SYSTEM

MAXIMUM
SYNCHRONOUS
FlATE

MAXIMUM
PARALLEL
RATE

DECSYSTEM-10

40.8 KB

lAS

19.2 KB

RSTS/E

19.2 KB

RSX-11 D

19.2 KB

125 KW

RSX-11 M

9.6 KB

40.8 KB

125 KW

RSX-11 S

9.6 KB

40.8 KB

125 KW

RTS-8

9.6 KB

RT11

9.6 KB

9.6 KB
TABLE A1

SUPPORTED DDCMP MODES FO'R DECNET SYSTEMS

SYSTEM

FULL
DUPLEX

DECSYSTEM-10

HALF
DUPLEX

POINT
TO
POINT

MULTIDROP

ASYNCHRONOUS

pYNCHRONOUS

PARALLEL

lAS

RSTS/E

RSX-11 D

RSX-11 M

X
X

RSX-11 S

RTS-8

RT11

X
X

TABLE A2

SUPPORTED NSP FEATURE:S FOR DEeNET SYSTEMS,

SYSTEM

LOGICAL
LINKS

INTERRUPT
FACILITY

MESSAGE
ROUTE-THRU

DYNAMIC
TOPOLOGY

DECSYSTEM-10

lAS

RSTS/E

RSX.-11 D

RSX-11 M

RSX-11 S

RTS-8

RT11

X
TABLE A3

SUPPORTED DAP FEATURES FOR DECNET SYSTEMS
DEVICE ACCESS
REMOTE
LOCAL

SYSTEM

FILE ACCESS
REMOTE
LOCAL

DECSYSTEM-10

lAS

X* *

X**

TSTS/E

RSX-'li D

X*"

RSX-'li M

X**

RSX-'li S

X**

VIRTUAL TERMINALS
LOCAL
R:EMOTE
X

RTS-B
RTii

'LOCAL' and 'REMOTE' are defined from the perspective of the application program. Thus, 'LOCAL' DAP support for
each feature implies the ability for the application program to access facilities on remote systems. 'REMOTE' DAP
support implies the ability to service DAP requests originating elsewhere in the network.
TABLE A4
* Operates on sequential files only,
* * Remote acces's only to device-independent characteristics,

DEeNET FUNCTIONS BY HOST ENVIRONMENT PAIRS
FUNCTION
INTH~-PROGRAM

SYSTEM
from

ALL

SYSTEM
to

ALL

with

DEC-10
lAS
RSTS/E
RSX-ii D
RSX-ii M
RSX-ii S
RTii

COMMUNICATION
INTEI=!-SYSTEM
DEVICE SHARING

DEC-i0
lAS
RSTS/E
RSX-11D
RSX-1i M
RSX-ii S
RT11

INTEI=!-SYSTEM

DOWN-LINE
SYSTEM LOADING

DOWN-LINE PROGRAM
LOADING

lAS
RSTS/E
RSX-1i D
RSX-i1 M
RT11

with

lAS
RSTS/E
RSX-ii D
RSX-i1 M
RSX-ii S
RTii

from

DEC-i0

DAS8S

from

RSX-ii D
RSX-iiM
lAS

RSX-1iM
RSX-iiS
RTi1

from

RSX-ii D
RSX-1i M
lAS

RSX-11 S

DOWN-LINE
PROGRAM COMMANDS

RSX-i1 D
RSX-ii M
RSX-11 S
lAS

from

RSX-1i D
RSX-1i M
RSX-11 S
lAS

INTER-SYSTEM
FILE TRANSFER

CROSS SYSTEM
SUPPORT

from

DEC-10
lAS
RSTS/E
RSX-11 D
RSX-11 M
RSX-11 S
RT11

from

RSX-11 M

DEC-10
lAS
RSTS/E
RSX-11 D
RSX-11 M
RSX-11 S
RT11

RSX-11 S

TABLE A5

DECNET INTE'RCONNEC:TABILITY MATRIX
DEC-10

lAS

RSX-11 D

RSTS/E

RSX-11 M

RSX-11 S

RT11

RtS-8

LEGEND:
DEC-10

S 40.8

~AS

S 19.2

11STS/E

S 19.2

I~SX-11 D

S 19.2

S XX-Synchronous interfa¢e,
maximum speed = XXKB
A XX-Asynchronous interMce,
maximum speed = XXKB
P XX-Parallel interface,
maximum speed = XXKW
-Systems not di rectly
interconnectable
<"
,:>

19.2

P 125
RSX-11M

S 40.8

S 19.2

S 19.2
p 125

S 40.8
A 9.6
P 125

I~SX-11 S

S 40.8

S 19.2

,:>

S 40.8
A 9.6
P 125

R> 125

9.6

A 9.6

9.6

S
A

9.6
9.6

A 9.6

S
A

9.6
9.6

19.2
P 125

rHS-8
f1T11

9.6

C'
\,)

9.6
9.6

$ 40.8
,of\

9.6

TABLE 1\6

DECNET PRODUCT CROSS-REFERENCE
PRODUCT

HOST EXECUTIVE

SPD REFERENCE*

DECNET-10

TOPS-10

8.60.0

DECNET /IAS

lAS

-10.71.0

SOFTWARE CODE**

OR680

DECNET/E

RSTS/E

-10.73.0

OP690

DECNET-11 D

RSX-11D

-10.70.0

OP680

DECNET-11 M

RSX-11M

10.75.0

OJ680

DECNET-11 S

RSX-11 S

-10.74.0

OJ690

DECNET/RTS

RTS-8

6. 1.0

OF680

DECNET/RT

RT11

-10.72.0

OJ685

TABU: A7
* Software Product Descriptions summarize the capabilities, features and required hardware for: each software package.
* * Software Code is the ordering number for the software package.