The official ECperf results for the Borland AppServer have been posted on ecperf.theserverside.com. The Borland results are the first ECperf results available. Hopefully the other vendors will follow soon. ECperf, composed of the Specification and the Kit, is a benchmark designed to measure performance and scalability of J2EE appservers.
Check out http://ecperf.theserverside.com
As I already said, good move... for now, if others don't follow, Borland (and their partners) can sell BAS/BES with the argument "The others don't even dare..." ;-)
I am glad that somebody does not afraid to show performance test results. And I am also happy that this somebody is Borland. Since my first experience with Borland's products (it was Turbo Pascal 6.0 for DOS) I never doubt in quality, performance and reliability of its software.
Looking forward when others will post their test results too.
Taras (Big fan of Borland)
What pleases me about this all is that we do not seem to have the same level of response (wars) shown when Oracle, IBM, BEA and others anounced there own little benchies.
This would imply there is support for ECperf in the community in providing accurate results which are reflective of production systems of the future (when people start moving more to EJB).
I applaud Borland (EBU/R&D) for being the first to come out with their results and congratulate them on what seem exceptional figures - SUN is even talking about them. Now all we need is some more results from the other appserver vendors.
I have been told that some vendors have been working hard on ECperf since the begining of this year. So we should expect some more results pretty soon.
Whatever the results I think I can now say without being attacked that Borlands AppServer is up there with the best.
Hehe, definitely true that Borland is currently top *gg*
I just wonder what it says about the ease of use of AppServers when more than one vendor works on ECPerf results and one comes out first with them. Lets see how much time passes until others come up with results.
I guess ECPerf is really accepted in the community, as it is really close to real world apps (opposed to e.g. TPC-C).
Also hope to see many more results from other vendors. Just hope that e.g. IBM does not "only" release results for their biggest (fastest) zSeries (the new Regatta I guess) as this would be somewhat meaningless. But I guess most vendors will show up with similar hardware (midrange servers 4-8 CPUs).
Congratulations to Borland!
But I can not see any results on http://ecperf.theserverside.com/ecperf
I always get the page saying no results has been posted yet.
What's up ?
You have to click on "Performance" under "Top 10 result by".
If you want to reproduce this yourself, we have made a toolkit available from :
Hope this helps,
This is great! I've used Borlands App. Server in the past, and I was definitely impressed with it. It doesn't surprise me that they were one of the first.
My guess is that Oracle and BEA will be next, simply due to the rivalry between them and throwing down of the gauntlet by Mr. Ellison to BEA and IBM earlier this year at JavaOne.
I'm also hoping that other non-mainstream vendors like JBoss and Orion participate and publish their results as well. It should be an interesting comparison, not to mention that it might give a bit more creditability to their cases when we try to convince IT managers their servers are the way to go. ;-)
Hmm, if Oracle publishes results you can expect Orion to be very similar as Oracle _is_ Orion.
If I remember correctly the last version I downloaded even was in the Orion packages)
--Posted by Bernhard Messerer 2001-11-01 14:08:40.0.
--Hmm, if Oracle publishes results you can expect Orion to ---be very similar as Oracle _is_ Orion.
--If I remember correctly the last version I downloaded ------even was in the Orion packages)
I may be wrong, but I expect that Oracle would use the full 9iAS functionality to run ECPerf metrics, and so I might expect that Oracle and Orion might come up with different results. Just a thought.
To say that ecPerf has been widely accepted by the community is pretty far-fetched given there has been only one submission to date. I find it hard to understand why J2EE app server vendors don't just simply use TPC-W. What are they afraid of?
It will be interesting to see other results if any get posted, but I question the value of a benchmark controlled by Sun in which only Sun-certified, J2EE-based products can compete. No vendor can submit a result if it does not pass Sun's certification...so that means no PHP, no .NET, and in fact nothing but J2EE-based products all based on the same core Sun technology can ever be compared using ecPerf. Talk about stifling innovation.
OK....flame me away.
PS: Check out the cost of the hardware used in the Borland submitted result. Wouldn't it truly be interesting if you could use PHP on Linux or .NET on Windows 2000 all running on Intel-based architecture, get better results, and cut the hardware cost by a factor of 5? But with ECPERF, we will never know.