Hello,

Does anyone have any useful way of providing a 'yardstick' for measuring the performance of the Baan application ? Are there any sessions which you run either on-line or batch which you feel are a valid and repeatable measure of the Baan applications performance ?

For example,

If we run MRP without any selection criteria, there is a temporary table dropped and recreated ( timrp010XXX ). At the start of the run, records are then written to this table during processing. At the end of the run we can calculate a rough throughput i.e. 7000records per minute ( based on MRP runtime and number of records written to this table ). The good thing about this is that we can run MRP in re-gen without refreshing data etc. Once we have the throughput figure we can then go onto another environment and check the throughput there.

This may simple but thats the good thing about it !

Any replies will be gratefully received !

Torwin,

You can search on the Board for the topics -

Profiler

ARM

Performance

You will find more information.

It is very difficult to generate a Baan yardstick. Things you can measure are:
- rows retrieved of a fixed table (eg. ttadv999) per second.
- Time of MRP during weekend diveded by the number of items.

But non of these will satisfy you.

The first is mostly datbase and hardware specific (a 200 MHz system will perform 2 times less compared to a 500 MHz system).

The second is dependend on:
- complexety of items.
- lots of parameters.
- HW, Configuration etc

As there are in general too many things that diffger between you and another company there is only one yard stick with is the Baan benchmark.

But, what did you want to do anyway with a yard stick? Improving performance on your machine? If that is the case, I think that NPRao made the best suggestions to search on.

Kind regards,
Dick

NPrao/Dick,

Thanks for the replies. I will certainly pursue the threads mentioned on BaanBoard. The reason I am asking about this particular method of measurement is that we are considering replacing our current hardware. We need to evaluate how our Baan application performs on this new hardware against our old stuff. I feel that if we go to the business and start talking about how much quicker the UNIX box is or how much better the Oracle database is running then they may not be interested. If we can tell them that we can process 14000 records a minute as opposed to 7000 records a minute then it will be simpler and more meaningful to them.

Looking forward to further comments....

A rule of thumb when comparing 2 systems with the same amount of CPU's is that the processing rates is the same according the TPC or Baan benchmarks. Of course this only counts when:
- disks are no bottle neck
- every thing runs on one machine.

In all other cases this rule of thumb becomes much more complex.

Kind regards,
Dick