Sap standard application Benchmarks

[antonio_voce]

Hi to all

We have an important question about Sap standard application Benchmarks

on the certification number 2007056.

In the certification number 2007056 is described the Central Server HP

Proliant DL580 G5 but is missing the description of the hard disk

structure used for the test : number of phisical hard disk and the speed

of the hard disk ( RPM 10k or 15K ) and the related chace.

Do you have more information regarding the structure and typology of thehard disk used for this Benchmark ?

Tank's.

Antonio Voce.

[markus_doehr2]

"Usually" you use the Quicksizer (http://service.sap.com/quicksizer) and as a result of it, you get a number of SAPS. Then you pass over that number to your hardware vendor (in this case HP) and they will size and propose an offer how to configure the system to achieve that number of SAPS.

Markus

[antonio_voce]

Markus

i know wath you tell.

The question is more specific , I want to know the number of the disk used for construct the array for the test.

You can link to this web site

http://download.sap.com/download.epd?context=40E2D9D5E00EEF7CE8AED38BFA3A8A0FFC1D711B956A4EA53514BC6...

you wil find all the information ( number of cpu , ram ) but NOT the number of the disk.

Sure if you have Operating system, SAP appserver, database and transaction log on 1 phisical hard drive , the number of mechanics (disk OPS ) is not sufficent to rich the number of the saps give in the document.

Any other suggestion ?

[markus_doehr2]

> Sure if you have Operating system, SAP appserver, database and transaction log on 1 phisical hard drive , the number of mechanics (disk OPS ) is not sufficent to rich the number of the saps give in the document.

I understand your questions

What you need to consider when using this benchmarks as reference is the fact, that the systems are tuned for those - which means: special database parameters that you wouldn't change in a normal environment (because you won't run into any bottlenecks on that area), specific disk layouts for those benchmarks, special operating system/network parameters etc.

I'm pretty sure, that even if you knew that all, you won't reach that number of SAPS on your system during production because benchmarks are synthetic and are based on assumptions (each order in VA01 has three line items always etc.) which will very unlikely be the same in your environment.

And what is the major thing using those: They are made for an avg. response time of almost 2 sec - which our (and your users) won't accept.

Consider them as "rough estimate", not as a "setup guideline" for a production system.

You can see how they are done (they are available) at

ftp://sapserv3/specific/benchmark

As you will see they are based on a fixed support package level, a special client etc. etc. - so they are not at all comparable to normal installations.

Markus