On an effort of continiously working to improve PostgreSQL performance with Solaris and Glassfish, it gives us great pleasure to announce the second SpecJAppServer2004 with PostgreSQL 8.2.4 and Glassfish V2 based Sun Java System Application Server 9.1 on Solaris 10. The improvements while for the database seems minor from 778.14 to 813.73 JOPS (which is like 4.5%), if you look at Glassfish itself, it now uses 33% less hardware and provided that 4.5% better performance which is an execellent feat in itself. Plus the CPU stress on PostgreSQL is also now reduced with optimization on how Glassfish v2 works with PostgreSQL. Its like music to CIO's ear, less hardware - more performance.
Before I forget to mention, PostgreSQL in the new benchmark is compiled with Sun Studio 12 with the following options:
-xO3 -xarch=v8 -xspace -W0,-Lt -W2,-Rcond_elim -Xa -xildoff -xc99=none -xCC
Why wait for others to do benchmarks, download and try Glassfish V2 and PostgreSQL 8.2.4 out yourself on Solaris 10/ Solaris Express Developer Edition of course !!
Disclosure Statement:
Sun Fire X4200 (4 chips, 8 cores) - 813.73 SPECjAppServer2004 JOPS@Standard
Sun Fire X4200 (6 chips, 12 cores) - 778.14 SPECjAppServer2004 JOPS@Standard
SPEC, SPECjAppServer reg tm of Standard Performance Evaluation Corporation. All results from www.spec.org as of July 23,2007
4 comments:
Nice to see. Not sure if that's something you can touch, but the url referenced should really be www.postgresql.org, not www.postgres.org. Yes, postgres.org also works, but postgresql.org is the official one.
This is really helpful for those, who are working on PostgreSQL. I am also working on OSDB PostgreSQL.
And I also doing Benchmarking (Performance metrics)
of PostgreSQL on Linux (Fedora).
Comparing the two benchmarks there is, as I see it, surprisingly little difference between the two tests. So I am interested in what was it that made such a big difference that you were able to remove one of the application servers and still improve performance?
The big difference between the two benchmark is the version of Glassfish aka Sun Java System Application Server. The earlier benchmark was based on Glassfish v1 aka Sun Java System Application Server v9. The later one was based on Glassfish v2 aka Sun Java System Application Server v9.1. The new Application Server is a huge improvment in terms of performance over the old one which resulted in 33% less hardware.
Post a Comment