(May 25, 2002) - An audit of the latest Oracle and Microsoft-published performance data for the Java and .NET Pet Shop found the .NET Pet Shop to be over ten times faster than the latest optimized J2EE Pet Store based on the latest Oracle-published benchmark data. The independent auditor, VeriTest, also found serious issues with the Oracle-revised Java Pet Store application and testing methodology, including missing application functionality and flawed benchmark load test settings.

This past March, Oracle published new benchmark data for the Java Pet Store based on a revised implementation of the Sun Java Pet Store 1.1.2. In May, VeriTest, a respected leader in independent software testing invited Oracle, to participate in an independent audit of their published benchmark data. Oracle declined to participate. VeriTest performed an audit of Oracle's data based on the revised Java Pet Store application and Mercury Interactive LoadRunner test scripts Oracle published on their Web site, and found serious flaws that raise questions about Oracle's latest published data.

VeriTest also performed extensive testing of the .Net Pet Shop to verify the original performance data taken using Quest Software Benchmark Factory, as well as performing new .NET tests with Mercury Interactive LoadRunner and Oracle's own published test scripts and data load. The tests were run on the same Compaq Computer Corporation ProLiant 8500 servers as used in the original Microsoft testing. The VeriTest audit shows that:

1. Oracle's latest published test data using Mercury Interactive LoadRunner is seriously flawed.
2. Microsoft's original published data using Quest Software Benchmark Factory is accurate but cannot be compared to results taken by Mercury Interactive LoadRunner due to differences in the load test tools.
3. When tested using Mercury Interactive LoadRunner using Oracle's own test scripts, the .NET Pet Shop performs over 10 times faster than Oracle's Java Pet Store based on Oracle's latest (questionable) published data.