This is mostly my day job, so those of a non-geek persuasion should look away now.
You may have noticed reports of a
Java versus .Net benchmark [caution: 2.2MB PDF file] doing the rounds, whach alleges that .net is faster than J2EE. Speaking as a professional J2EE person, if what
this article says is true, the benchmark is seriously flawed, to the point where I start to suspect deliberate bias. Of course, M$ is trumpeting this result to the moon and back.
Now, I'll freely admit that J2EE ain't the fastest architecture in the world, but that's not its primary aim. J2EE is mostly aimed at reducing total cost of ownership, by encouraging (with extreme prejudice) several forms of "best practice" in the design and programming of systems, and allowing code to work cross-platform. It sounds like the .net system tested here is seriously flawed in the "best practice" respect. To add to the injustice, having played with the J2EE petstore, it's a combination of technology demonstrator and example code for teaching purposes, not any kind of performance benchmark tool. They're comparing a "real car" to a cutaway "how a car works" model. Apples and oranges.
Which would you rather have, a system that runs 10% faster, or a system that is well designed and easier to debug?
[links from
The Register]
[edit: tidy up, fix the typos, make mental note to copyedit more when drunk]
