UNIX TruCluster Available AlphaServer 4x00 DIGITAL

This document, published by Digital Equipment Corporation in March 1997, provides an overview and detailed technical information about the interoperability test results for an Internet UNIX TruCluster Available Server 4000/4100 running DIGITAL UNIX Version 4.0b.

The document serves as a guide for configuring prequalified computer systems (HiTest Systems) for medium to high-end UNIX users, ensuring high system reliability, application performance, and upgradability. It details the specific hardware, software, and firmware components that were tested together, outlining both minimum and maximum configurations, including AlphaServer 4000 and 4100 models with one or two CPUs and 512MB to 2GB of memory.

Key aspects covered include:

• Configuration Data: Tables listing specific part numbers and tested ranges for foundation hardware (e.g., AlphaServer 4000/4100 base systems, memory, storage, CPUs) and software (DIGITAL UNIX V4.0b, ServerWORKS, StorageWorks, Advance File System, LSM, Networker). Special configuration rules are also provided, such as unique SCSI host IDs and specific memory slot assignments.
• Installation and Setup: Guidelines for software installation, emphasizing the use of AdvFS for file systems and proper licensing, followed by TruCluster Available Server setup and application installation.
• Interoperability Tests and Results: Describes testing using a database application workload, driven by an automated tool, to verify data integrity under various failure scenarios. Both minimum (500KB database, 1-day testing) and maximum (24GB database, 5-day testing) configurations were tested for routine shutdowns, power failures, machine halts/panics, shared disk/network disconnects, hotswapping, and storage shelf failures. Load testing also included NetWorker backups/restores and DIGITAL Internet AlphaServer (IAS) traffic. All tests successfully demonstrated the TruCluster Available Server's ability to perform failovers and maintain service.
• System Limits and Characterization Data: Notes that failover is not instantaneous and depends on the type of failure and application startup time. It also highlights that system availability is contingent on redundancy levels and that response times increase with higher application loads and user counts.
• Problems and Resolutions: Addresses common issues encountered during improper installation or configuration, such as kernel build errors, boot failures due to missing name servers, inaccessible shared disks, asemgr hangs, and problems with altering member network interfaces or disk services disappearing.

The document is intended for sales representatives, technical support, customers, product managers, and personnel responsible for HiTest System installation and operation.