(Story 1) At a customer site they upgraded to new IBM machines, increasing capacity by 2,000 MIPS; more LPARs were concentrated on the same machine and the performance was worse than before; EPV highlighted the PRSM overhead due to high logical to physical CP ratio; the customer reduced the number of logical CPs assigned to each LPAR and got back more than 1,000 MIPS and the expected performance.

(Story 2) EPV highlighted a high page fault rate on a production system impacting CICS applications. Performance Analysts looking at the memory assigned to each LPAR in the machine discovered 16 GB offline to all LPARs which were once dedicated to another production system (migrated to a different CEC); they used part of the “lost” memory to reduce the page fault rate to zero and improve application performance.