Designing an architecture is one thing, believing that it will deliver
the expected performance without any hard figures another.
Performance is always excellent during unitary tests, and often adequate after a number of adjustments even when testing against true production volumes.
However, the effects of concurrency, contention, serialization and what is generally known as thrashing are notoriously hard to quantify before going live.
Doing a serious benchmark is a long and technically challenging operation which requires highly skilled specialists, as much at the definition as at the execution level.
RoughSea teams up for benchmarks with Alcane, a specialist French
company made of people also deeply rooted in the Oracle world. Togther we can help you set up and run benchmarks to give you full confidence before going live.