This chapter covers key OpenAM tunings to ensure smoothly performing access and federation management services, and to maximize throughput while minimizing response times.

Note

The recommendations provided here are guidelines for your testing rather than hard and fast rules for every situation. Said another way, the fact that a given setting is configurable implies that no one setting is right in all circumstances. The extent to which performance tuning advice applies depends to a large extent on your requirements, on your workload, and on what resources you have available. Test suggestions before rolling them out into production.

As a rule of thumb, an OpenAM server in production with a 3 GB heap can handle 100,000 sessions. Although you might be tempted to use a larger heap with a 64-bit JVM, smaller heaps are easier to manage. Thus, rather than scaling single servers up to increase the total number of simultaneous sessions, consider scaling out by adding more servers instead. The suggestions that follow pertain to production servers.