Oracle has announced world record multi-node SPECjAppServer2004 benchmark results for Oracle Fusion Middleware. The combination of Oracle WebLogic Server, a component of Oracle Fusion Middleware, together with Oracle Database 11g on mission-critical HP Integrity server blades running HP-UX 11i v3 delivered the record-breaking results.
SPECjAppServer2004 is a multi-tier benchmark for measuring the performance of a representative J2EE application and each of the components that make up the application environment, including hardware, application server software, JVM software, database software, JDBC drivers and the system network.
"Certainly first and foremost, it is a world record," Mike Piech, senior director of product marketing for the application server and related products within Oracle Fusion, tells 5 Minute Briefing, commenting on the significance of the benchmark. It marks the "unquestionable top performance in this category," he adds.
"In fact, we actually are the leader in all of the SPECjAppServer categories, but in this one, in particular, which is multi-node, we are number one by a substantial margin.
A key point, notes Piech is that "we have achieved this number-one spot using fewer cores than the holder of the number two spot." According to Piech, "if you do the math and look at it, on a per core basis, our score is about 28 percent higher in terms of performance. That is not eking out a small increment over the next guy, neck and neck."
Oracle WebLogic Server 10g R 3 together with Oracle Database 11g achieved 21,602.31 SPECjAppServer2004 JOPS@Standard (jAppServer Operations Per Second). Oracle WebLogic Server was running on 13 HP Integrity BL870c server blades, each with four Dual-Core Intel Itanium 1.6 GHz processors. The database server was running on an HP Integrity Superdome server with 64 Dual-Core Intel® Itanium 1.6 GHz processors and HP StorageWorks 8100 Enterprise Virtual Arrays. Both application server and database tiers were running HP-UX 11i v3.
"SPECjAppServer is a really unique benchmark in that it very well represents a typical workload," Piech observes. "Some kinds of benchmarks are very simplistic raw performance numbers that don't, at the end of the day, translate to a metric that is all that useful when trying to get a sense of how something is going to work in a real-world application," he explains. However, "SPECjAppServer is unique in representing real-world workloads so this is a very relevant benchmark for our customers." For more information on SPECjAppServer2004, go here. For more information on Oracle benchmarks, go here.