Oracle benchmarks

Oracle Database Tips by Donald Burleson

For the IT manager charged with predicting Oracle database performance, benchmark studies present a huge challenge. It's very difficult to use Oracle benchmarks to "prove" that a artificial load is representative of real-world processing conditions.

There has been a backlash against the test-case scenarios where a small and inconsequential test is extrapolated to a larger system, and today's Oracle professionals recognize that there is no short-cut for a performance simulation that closely resembles real-world processing.  As user load increases, data concurrency issues arise and we must be able to forecast Oracle bottlenecks that arise as a function of high user activity.

Simulating the activities of a large number of end-users is a giant challenge, but there are some important tips and tools used by benchmarking professionals to ensure that we develop a statistically legitimate model that will predict actual user loads.

I highly recommend Bert Scalzo's new book on Database Benchmarking book for Oracle and SQL Server for details on conducting valid Oracle database benchmarks.  My other notes on Oracle benchmarks include: