Re: Oracle Benchmark Results for Different Hardware Configurations?

"HansF" <Fuzzy.Greybeard_at_gmail.com> wrote in message news:pan.2006.09.14.13.24.19.798555_at_gmail.com...
> On Thu, 14 Sep 2006 03:48:49 +0000, Bob Jones wrote:
>
>>
>> How does TPC results reflect the performance of a particular Siebel or
>> SAP
>> implementation?
>>
>
> Nope. And it shouldn't. That is perhaps why there are several different
> benchmarks in the TPC studies, each reflecting different workloads.
> Resulting in the ability for people to make a reasonable guess.
>

If they do not reflect the real performance, how does one arrive at a "reasonable guess"?

>>
>> Again, what's the use of TPC if you are doing your own benchmarking?
>
> <sigh>
> TPC is a reference. It is not intended to reflect a specific application,
> but to give a way of comparing things through ratios ... in spite of all
> the problems with ratios.
>
> Benchmarking the actual application *properly* is always preferable to
> TPC (unless the benchmark must take politics into account). Like all the
> other public benchmarks, it is designed as a reference and must be used
> intelligently.
>
>

How can it be a reference if it does not reflect the real performance at all?

>
> From your stance on this subject, I take it you (or your organization) is
> fortunate enough to have the skills, time, equipment and vendor
> relationships to be able to properly benchmark - /including verifiably
> correct application, OS and hardware tuning/ - for every application,
> server, disk and I/O controller combination of interest, at the
> application full load run time configuration. For that I commend you ....
>

Quite the contrary, I don't even have 15 people and 3 months like you did.

> No need to drag this on - you get the last comment.
>

My last comment is that TPC can be misleading and rarely usful for real world performance plannings. Received on Thu Sep 14 2006 - 19:00:19 CDT