Pet Shop 2.0: Java vs. .NET

Here. The Middleware Company ran a comparison between the Pet Shop v2.0 application for both Java and .NET, both with the same features and both with the same aim at building something to use as a "best practice" sample while still being as responsive and scalable as possible. Pet Shop v2.0 was both company's attempt at winning the comparision run by the Middleware Company. My reading of this report is that .NET kicked Java's hinder in every single measure, from through-put and responsiveness to lines of code and lines of configuration required to build and run the app.

Interesting. However, I can't seem to find the section where they replace the poorly performing and expensive J2EE servers (probably BEA and Iona) with Open Source JBoss, thereby reducing the J2EE costs to well below .Net, thus soundly kicking its rear.

Also, no mention is made of the pricey dev tools needed for .Net. Wouldn't it be fair to add them in?

cheers

While performance may be important for a certain class of applications (the BIG ones), I think development speed, ease of maintenance etc are much more important for many others (spend a week of development time optimizing your code, or spend $500 on more RAM). Fortunately .NET is great for developer productivity. You only have to look at the KLOC count for both implementations to realise the benefits in this area that .NET brings.

.NET beat J2EE on price *and* performance. Replacing the expensive commerical J2EE server with JBoss won't make the system run faster. (Are there any JBoss official perf benchmarks? Not according to their site.) Also, since .NET runtime has no cost, JBoss would bring the J2EE system in line with the .NET servers - not exactly "soundly kicking its rear".

As for tools, the .NET SDK is a free download, and there are a variety of free tools (ASP.NET Web Matrix, SharpDevelop) that can be used to develop for .NET. Visual Studio .NET is more price, but it's comperably priced to commercial J2EE vendor tools.

This study shows .NET is much easier, much faster and much cheaper than J2EE. Well, with JBoss, I guess it's only much faster and much easier.

Actually, JBoss outperforms the other J2EE app servers quite substantially (there are a number of metrics around the web) bringing the J2EE performance much closer to .Net. Not faster, I won't argue that, since .Net is bound to a single platform and hence is optimized for that single platform. J2EE will always pay a price for the freedom to choose a platform. Usually a more secure platform.

So, the .Net advantage remains only when both platforms run on the same OS (Windows). When you run JBoss on Linux, JBoss will obviously get the better metric (price/performance) since for Windows you still have to pay for the OS. Simple maths, Dr Watson.

As for ease of use (deployment descriptors etc) we're talking very trivial issues. Too trivial to really be a factor.

cheers

I agree with most of GingerJ said.

As far as the report itself is concerned, take a look at http://dreambean.com/petstore.html.

It's from a JBoss guy, said something like "while the J2EE PetStore code goes through some serious hoops to
store parts of a category product list in the user session, the .Net code
simply caches the entire category product listing in an application-level
cache."

I must say it amazes me that people in the software industry put any weight behind these kinds of 'impartial reports', either from MS, Sun or any other company. Its purely marketing material.

I would be interested in comments agreeing with or disagreeing with this rebuttal...

http://dreambean.com/petstore.html

He brought up several issues however no matter which way he twisted, J2EE still lost.