California Governor Arnold Schwarzenegger will join Oracle CEO Larry Ellison today at Oracle OpenWorld. I can fantasize that as Arnold gets up on stage he "calls out" Larry on his benchmarks like he did with his wife recently on her cell phone use while driving.
Oracle once again has made invalid benchmark claims in a press release -- this time on SPECjAppServer.
- Oracle utilized a 48 core server in this benchmark, not the 6-core one their press release leads you to believe. IBM WebSphere Application Server (WAS) was benchmarked on a 4 core server. Note that while Oracle reports 7x the performance of IBM it was with 12x the number of processor cores (1).
- The IBM WAS benchmark result used as a comparison was conducted in August 2008, 14 months ago and utilized WAS v6.1. WAS v7.0 was made available in September 2008 and provides superior performance to WAS v6.1.
- Oracle claims they are the leader in multiple SPECjAppServer benchmark categories - but they have gone back to 8 year old retired benchmarks which have no relevance to today's IT infrastructure - SPECjAppServer2001, SPECjAppServer2002, and even, can you believe it, ECperf. Those were the days.
- The most recent SPECjAppServer2004 benchmark result from IBM using WAS v7.0, released in January 2009, compared with an Oracle WebLogic Server result using the same amount of application processor cores and the same software version as cited in the release clearly shows IBM WAS v7.0 as the dominate price/performance leader (2).
The fantasy continues as Arnold carries Larry off the stage with an "I'll be back."
(1) Oracle Web Logic Server on HP ProLiant DL785 G6 with a Six-Core AMD Opteron 8439 SE 2.8 GHz processor (48 cores total), 9,455.17 SPECjAppServer2004 JOPS@ Standard. IBM WebSphere 6.1 Application Server on IBM System p570 with 2 processors, 4 cores, IBM POWER6 4.7 GHz, 1,197.51 SPECjAppServer2004 JOPS@Standard.
(2) Based on dividing the list price for the application server software used in the benchmark by the throughput of the benchmark result for Oracle (http://www.spec.org/jAppServer2004/results/res2009q2/jAppServer2004-20090325-00130.html) and IBM (http://www.spec.org/jAppServer2004/results/res2008q4/jAppServer2004-20081216-00126.html) on results with comparable throughput levels.
Source: http://www.spec.org; Results current as of 10/14/09.
SPEC, SPECint, SPECfp, SPECjbb, SPECweb, SPECjAppServer, SPECompM, SPECompL, SPECsfs, SPECpower, SPEC MPI and SPECpower_ssj are trademarks of the Standard Performance Evaluation Corporation (SPEC).
The postings on this site solely reflect the personal views of the author and do not necessarily represent the views, positions, strategies or opinions of IBM or IBM management.
technorati tags: IBM, systems, performance,