Why are many performance tests considered failures?
A performance test is one of the most complex tests to plan and design properly. I have found that there are several key areas that affect the usefulness of the test and its results.
1. Starting the testing too late.
2. Architecture issues: If the physical layout of the test system is substantially different from the production or target system, the test may not provide useful performance data.
3. Confusing business goals with performance goals.
For more elaborate answer, see Better Software Magazine [November/December 2010 issue].