Originally posted by: SystemAdmin

Hi,

Why IBM uses System P servers in TPC benchmarks and not mainframe systems (System Z)? Does it mean that high-end P machines have better performance than Z machins? I guess that mainframes are used for near-zero down time application, and probably big unix servers have better performance than Z machines. Am I alright?

Thanks alot.

Originally posted by: MarkTaylor

yes, system p are faster than system z for TPC .. p == performance and as you quite rightly pointed out z == zero down time .. although, nowadays all the system z technology is filtering down to p & i, and in the future it will all run on the same h/w anyways so the limiting factor will be the o/s.

HTH
Mark Taylor

Originally posted by: SystemAdmin

At the risk of putting words in their mouth I'd say that IBM runs TPC benchmarks on pSeries because customers can run AIX workloads on other UNIX servers fairly easily. That makes it a competitive thing, IBM has to demonstrate how much faster their boxes are than the other guys. Mainframes are more of a captive market and don't substitute directly for anything else so IBM only needs to compare their zSeries models to each other rather than anything else. It's also a historical thing. TPC started life in the "open systems" market. Mainframes have their own measurement schemes, MIPS and all that.

HTH

Jim Lane