Platform (integration) testing by IBM

For each release of z/OS®, a team of IBM® testers and system programmers performs a IBM Z® platform test (formerly known as integration test). Test systems simulate customer production Parallel Sysplex® environments running z/OS and many related software products such as Db2®, IMS, CICS®, IBM MQ, and WebSphere® Application Server. The team migrates to each new release of z/OS, puts the system into production, and then verifies that all products applicable to each operating system work together. The team not only tests software, but also verifies that the software runs in a hardware environment that has both generally available (GA) and pre-GA levels of IBM Z® hardware.

The zPET team maintains two parallel sysplex environments. They have a 4-way test and quality assurance sysplex, and a 9-way production sysplex. Like most customers, the latest z/OS gets migrated by LPAR first to the 4-way system and later to the 9-way environment. Migration and coexistence are tested between the pre-GA z/OS code (n–1) and the latest z/OS GA code (n). After fully migrated to the pre-GA z/OS, a fallback test scenario is performed. If successful, the fallback LPAR is once again migrated to pre-GA z/OS and tests of new functions commence.

The team documents their experiences with installing, migrating, and testing z/OS and IBM Z® hardware and System Storage®, in LCST/e z Systems Platform Evaluation Test Community blog, which is available from IBM z Systems Platform Test. This website also includes test results and advice for customers, and illustrations of the team's hardware, software, networking, and application enablement configurations.

The team follows a process that includes simulating the workload stress levels, transaction rates, and lock contention rates that customers might experience. They stress many of the same areas of the system that customers stress. In their report, they include detailed descriptions of their hardware, software, networking, and application enablement configurations, as well as the key operational, performance, and availability characteristics of their Parallel Sysplex environment. They provide recommended recovery procedures and various hints and tips that are distilled from their own experiences. They explain the mistakes that they make so that you can avoid those mistakes.