Blogroll Me!

Lives in India/Chennai/Chennai, speaks English.
This is my blogchalk:
India, Chennai, Chennai, English.

A blog on what is happening in the .NET world in India.  

Anand's .NET WebLog


Home Archives Contact

Thursday, July 31, 2003 :::
New TMC Benchmarks:::  

Robert A. Wlodarczyk points to the new TMC benchmarks of .NET vs. Java. The first benchmark showed that .NET was 2-3x faster that Java and this resulted in a lot of critisisms form the Java camp. TMC has now redone the whole thing with the involvement of the Java community etc. The report is avbl at the TMC Site.

This time Java has performed better and in some cases the performances are almost the same, while in others .NET still maintains the lead.

Points to Note: (from Robert's post)


Test 1: Web Application Throughput with Oracle 9i:

.NET (peak throughput of web pages/sec) : 1586.54
J2EE-EJB-CMP (peak throughput of web pages/sec) : 1585.74 (for app server X)
J2EE-SERVLET-JSP (peak throughput of web pages/sec) : 1094.00 (for app server X)
(I don't even [re]mention app server Y since the results are, well, pitiful.)
Test 2: Web Application Throughput with SQL Server 2000:

.NET (peak throughput of web pages/sec) : 1773.14
J2EE-EJB-CMP (peak throughput of web pages/sec) : 1578.20 (for app server X)
J2EE-SERVLET-JSP (peak throughput of web pages/sec) : 1003.81 (for app server X)
(I don't even [re]mention app server Y since the results are, well, pitiful.)
Test 3: Web Services Throughput:

.NET & Oracle (peak throughput of web service requests/sec) : 832
.NET & SQL Server 2000 (peak throughput of web service requests/sec) : 1245
J2EE-EJB-CMP (peak throughput of web service requests/sec) : 312 (for app server X)
J2EE-SERVLET-JSP (peak throughput of web service requests/sec) : 359 (for app server X)
(I don't even [re]mention app server Y since the results are, well, pitiful.)
So, from these results, we can learn a few things:

After 9 months+ since the previous TMC Performance Benchmark, the Java folks have had time to bitch and complain, tune, re-tune, re-re-tune, and they reach performance similar to that of .NET (only in the Web Application market). So, lesson here: if you want speed of deployment and great performance, .NET is the wiser choice. Less time to market. Period.
Still after 9 months+ time, neither app server X or Y are able to compete with .NET. .NET wins hands down. They aren't even close. Period. Even when .NET is used with Oracle for Web Services, it still beats Java solutions. So, I think the myth of Java-based Web Services is shown clearly here... it's still a myth.
I know this will probably start a something close to a war, but seriously, the Java Community had their chance this time around to tweak, complain, or what-have-you, and still .NET comes out on tops when it comes to Web Services... and in the Web Application market, .NET is still a faster means to market.




::: posted by Anand at 12:40 PM

Comments: Post a Comment

 
Subscribe to Anand's .NET WebLog!


powered by Bloglet


Listed on Blogwise Listed on BlogShares Technorati Profile

Powered by Blogger