HP has published another SPECjAppServer2004 J2EE benchmark using WebLogic Server 9 scoring 3,734.68 JOPS@Standard. This beats the previous highest score of 3,328.80 JOPS@Standard, also on WebLogic Server 9, posted in November 2005 by Sun Microsystems.
- Posted by: Gary McBride
- Posted on: January 05 2006 16:28 EST
Since its general availability in July 2005, WebLogic Server 9, on HP and Sun hardware, has posted 5 world record scores on the SPECjAppServer2004 benchmark, holding the top spot from then until now all but 4 weeks.
As of January 5th, 2006, there are a total of 16 SPECjAppServer2004 results to compare. Seven of them use WebLogic Server 9.0, five use WebSphere 5/6 and four use Sun Java System Application Server.
SPECjAppServer is a trademark of the Standard Performance Evaluation Corp. (SPEC). Competitive numbers shown reflect results published on www.spec.org as of January 5th.
It would be nice to see some numbers for JBoss and Geronimo. I know it takes money to set these things up and run them, but some comparison would be helpful (i.e. is Geronimo 5x slower than WebLogic?) especially in the $/work unit metric.
I think with Geronimo you can do better than just measure overall performance. Each sub-project can measure its maximum throughput separately, which is great for predicting performance before any of the code is written.
Enjoy the Fastest Known Totally-Ordered Reliable Multicast Protocol
Too bad there's not a benchmark for evaluating end user code performance, since 9 times out of 10, most application performance issues I see are due to developer's code and not the application server performance. It would be nice to say "I have code that scores a 5,234 COPS@Standard with the SPECjAppCode2005 benchmark". That way you could easily track down developers whose code is not up to snuff :)
That is true. Margin of performance difference of application server is very small. If we can calculate the rate of J2EE developer or consultant base upon the "COPS@Standard with the SPECjAppCode2005 benchmark", it would be great. I really do not like that people complain slowness of Java but writing very lousy Java code. For sake of J2EE programming, we need train more good J2EE developer instead of improving small margin of performance of Application Server.