|
|
Posted by DA Morgan on 04/17/06 18:46
Erland Sommarskog wrote:
> Tony Rogerson (tonyrogerson@sqlserverfaq.com) writes:
>> There is no point, I'd rather trust the TPC for independence instead of
>> what you will produce ie. a biased benchmark based on your
>> anti-microsoft stance and lack of technical ability with that platform.
>>
>> Like I say, check out the benchmark on tpc.org.
>>
>> On checking TPC there is a comparitive benchmark where SQL Server beats
>> Oracle hands down on the same hardware (HP Integrity Superdome), SQL
>> Server -> 1.2million; Oracle -> 1million tpmC
>> (http://tpc.org/tpcc/results/tpcc_perf_results.asp) - no bias there!
>
> I think I agree with DA here. TPC is good if all you want to do is
> size-comparisons of male organs, but if you want to test for your actual
> application, you should run your own benchmark. And keep in mind that
> that benchmark applies to that application, and necessarily not any other
> application.
>
> As I understood it, DA's benchmark is for the same Oracle application
> on Windows and Linux, which should be a fairly trivial benchmark to run.
> (But it may say more about Oracle's implementation on the two operating
> systems, that it tells about the operating systems themselves.)
>
> Running a benchmark for the same application running SQL Server and
> Oracle is a far more devilish game.
Precisely. Because then you are dealing with SQL*Net vs another
connection mechanism. You are dealing with array processing versus
cursors, and lots of other real-world compromises we all make when
we decide whether to implement best-of-breed functionality or try
to stay generic. I think Tony, in his zeal to defend turf, missed
the point I was trying to make. Next time I will try to be clearer.
--
Daniel A. Morgan
http://www.psoug.org
damorgan@x.washington.edu
(replace x with u to respond)
Navigation:
[Reply to this message]
|