c0t0d0s0.org

When you look at the recent SAP SD certifications about the HP and the Sun systems based on Intels new hexacore proc, a cursory observer may have the impression, that either Solaris10+MaxDB is much slower than Windows+SQLServer or that Sun has somehow has fscked up the System. But as usual the explanation is in the details: The HP/Microsoft result is a non-unicode result, the Sun result is a unicode result. Why is this important? There is a good presentation made by SAP about the expected performance impact: between 15% more CPU load on new procs and 30% more CPU load on older procs and 50% more memory by using Unicode. So it´s important to know for reading benchmarketing: You can´t compare non-unicode and unicode benchmark results for SAP.

There are some interesting benchmark numbers in the press release from Sun about the new SPARC Enterprise M9000 with 64 SPARC VII ("Jupiter") processors:

The SPARC Enterprise M9000 server running Solaris 10 OS, the SAP ERP application Release 6.0 and Oracle Database 10g achieved a world record result on the SAP Sales and Distribution (SD) Standard Application Benchmark by achieving 39,100 SD Benchmark users. The SPARC Enterprise M9000 server result beats the IBM Power 595 POWER6 server with 35,400 SD Benchmark users by 10.5% and outperforms the HP Integrity Superdome SD64B with 30,000 SD Benchmark users by 30.3%,[...mabushi deleted...].[1]

In addition to that, Sun announces some impressive numbers for the LINPACK benchmark. 2 Teraflops in a single operating system image

The SPARC Enterprise M9000 server, based on SPARC64 VII processors, and using Sun Studio 12 software with Solaris 10 OS delivers a score of 2.023 TFLOPS on the Linpack's Highly Parallel Computing benchmark. The SPARC Enterprise M9000 outperforms the IBM Power 595 POWER6 by nearly 2X and beats the HP Integrity Superdome system by 2.7X.[2]

This is interesting for computing jobs in need of a large single system image with short ways to to the memory. (No, the SGI Altix isn´t a competition for that ... single image too, but completely different architecture). Not all jobs scales on a cluster machine

Mithun from Sun ISV engineering made some very interesting studies about the performance of SAP on Xen. It´s a very interesting read and gives you some good insights into the challenges of virtualising I/O heavy tasks. He published a draft of his whitepaper on his blog.

Wow ... the Sun T5120 is the first single socket system to break the 10.000 SAPS barrier (you will find the certifcate here) ...

"Creating SAP system clones using Solaris 10 virtualization concepts" written by Mark Wondratschek of SAP shows, how you can use zones in conjunction to zfs to clone your SAP installations. The concept is quite handy for tests, updates and patches. The tutorial consists out of four parts, so be sure to follow the links at the end of each text.