Benchmarking and systems performance - hosted by Elisabeth Stahl

Elisabeth Stahl

Blog Authors:  ELISABETH STAHL   is Chief Technical Strategist and Executive IT Specialist, IBM Systems and Technology Group, and has been working in systems performance for over 25 years.

  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

Comments (1)

1 David_Davidian commented Permalink

Sun spent nearly a decade claiming TPC-C doesn't represent any real data center application (even though there have been internal Sun white papers challenging that claim) and moreover, tpmC results cannot be used for direct system sizing. The latter can be claimed for most large system benchmarks and is nothing more than adding fuel to Sun avoiding publishing TPC-C results. Of course, this neglects the relative and comparative nature of these types of benchmarks. The same argument can be made for another benchmark, SPECjbb2005, yet Sun publishes its results and touts world record results.<div>&nbsp;</div> It is probably not surprising that Sun began its anti-TPC-C rampage as the UltraSPARC-III was being introduced in late 1999/early 2000. Until that time apparently there was no problem with TPC-C, considering Sun’s E10k was pumping out world record results.<div>&nbsp;</div> Due to the numerous, yet thin threaded nature of the TPC-C benchmark, Sun's Niagara family of processors actually do well on TPC-C, given adequate I/O. But because of years of badmouthing the benchmark, Sun painted itself into a corner and could not publish those results without appearing extremely duplicitous.<div>&nbsp;</div> Now it appears, with Oracle’s insistence, Oracle+Sun may attempt to publish results on this “previously worthless” benchmark. Reading between the lines, Sun+Oracle might publish a result so high as to make a mockery of the fundamental usefulness of this benchmark. To some this appears as a vendetta by Sun to make up for nearly a decade of not having TPC-C results upon customer request, but instead generated excuses why they don’t have any published results.<div>&nbsp;</div> One would have expected Sun to simply ignore this benchmark rather than to engage in revenge marketing.<div>&nbsp;</div> -David Davidian dbdavidi@us.ibm.com

Add a Comment Add a Comment