CHAPTeR 9 DXP PeRfORMANCe OPTIMIzATION • Recommendations: • Minimize session size by storing only absolutely needed objects in the session. • Size the infrastructure to handle the maximum user load. • Use multinode clustered topology for all servers to eliminate single point of failures. A multinode topology involves using multiple nodes in a setup to handle user requests. • Use a caching framework or caching server (a dedicated server to cache frequently used objects) to cache frequently used data and lookup data. • Configure the application server parameters to handle maximum user load. The common parameters for Java-based application servers are given below: • Garbage collection metrics • Tuning parameters • Xms, Xmx • NewSize, MaxNewSize • PermGen, MaxPermGen • Performance issue: Inappropriate deployment architecture leading to single point of failure. Few common symptoms given below: • Multiple applications were deployed on the same server in production and pre-production environments. Performance testing revealed that this setup was not scalable. • CPU utilization graphs showed spikes during normal load testing. This was attributed to problems with one of the applications, which was not scalable. • Applications, admin functionality, and search modules shared the same server. This led to performance issues in search and admin functionality. 253