Re: TCP benchmarks

tolly brumbelow wrote: > > Snorri Bergmann <snorri@strengur.is> wrote: > >[Stuff deleted about the benchmarks ] [First, please ignore spelling errors. Netscape should really have a spell checker.] Take the benchmarks with a grain of salt. TPC/C is supposed to be able to reduce a vendor from tweaking their product to produce unobtainable results. However, if you look at the Oracle ads, and even the Informix ads, you will find some disturbing truths. 1) Products used in benchmarks are not yet released to general public. Oracle has 7.3 Informix 7.2 2) Most ads pit their best numbers against an other vendor's numbers even though the hardware is different and the costs are skewed. Informix's last add compares its products versus Sybase and Oracle on the same hardware. (Only Informix used 7.2) The bottom line you need to run independant tests on both the hardware and software. But beware.Both hardware and software vendors will still try and tweak the results in their favor. Don't put too much faith in the numbers you see in print. -Mike #include <std.disclaimer>