(November 20, 2002) - Late last month, the Middleware Company published a report called "The Petstore Revisited: J2EE vs. .NET Application Server Performance Benchmark." This report marked the first time that a non-vendor and J2EE expert has attempted to objectively compare and contrast the performance characteristics of .NET with those of Java 2 Enterprise Edition (J2EE).

The tests were conducted by the Middleware Company at Microsoft's labs, following Microsoft's acceptance of the Middleware Company's challenge to prove their earlier claims about .NET performance. To keep the playing field level, Sun Microsystems' own Pet Store demo was used for the benchmarking work - exactly as it had been used in previous attempts.

The results have been quite surprising to the J2EE community! First, .NET managed to beat BEA's WebLogic Server (referred to as Application Server "A" in the report, owing to licensing restrictions on disclosing performance metrics). Then, it went on to thoroughly defeat IBM's WebSphere server (referred to as Application Server "B"). Needless to say, the many throngs of Java supporters have not been happy to see this report released!

Here at the .NET Developer's Journal, we tend towards the view that all benchmarking tests are less-than-completely-accurate representations of the full depth and breadth of differences between platforms. This is as true on the rare occasion that a Java company publishes a report making .NET look good as it is on the frequent occasions when they publish faulty reports that make .NET look bad. So, in the truest, "decide-for-yourself" spirit of modern journalism, here are some links for you to investigate the matter further on your own:

www.middleware-company.com/j2eedotnetbench/

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bdasamppet.asp

www2.theserverside.com/home/thread.jsp?thread_id=16149