SAP PI 7.31 performance issues and high throughput

[Farid]

Hello,

We are running SAP 7.31 SPS4 Dual Stack, on Oracle11/HPUX

As the SAP Basis Administrator, I have been asked to investigate some perfroamce issues that occur every week-end,

when a particular interface is running (Idoc to AENGINE, as far as I can understand it)

I am working with the PI Perfromance document and tryning to walk though the troobleshooting procedure before contacting the PI developer.

http://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/70c8bdb7-6fe0-2e10-5b86-a33fdb53f...

So far, I have determined that the perfromance bottleneck occurs in the Integration Engine during the PLSRV_MAPPING_REQUEST step.

There is a dedicated chapter to the PLSRV_MAPPING_REQUEST step in the SAP PI Perfromance document, and I am still trying to determine if it is a system problem ( and therefore my problem)  or a problem with the implementation of the mapping of that specific interface

But right now, I am a bit concerned by the volume of data to be processed over a couple of hours :

Over the past week-end we have reach a throughtput of 88GB/hour,

My question may sound silly, but I have no references with others PI systems and I could not find any PI 7.3 benchmark (http://scn.sap.com/thread/3277249)

In your opinion, is this value too high for PI 7.3?

In others words, is it a problem wiht PI itself?

Thanks and Regards

[smavachee · Accepted solution]

Certainly it's at higher end, though it lot depend on message size and complexity of mapping used.

Please refer Page no 72 of the same document (Showing 10-12GB/h). Though its generic performance analysis indicator, you may need to work on it.

Keep watch on Memory / CPU utilization during peak hours, that would be more critical and it may cause Server Restart.

Regards,

Sunil

[rajasekhar_reddy14]

Definetly too high.

Check with PI consultant , he will provide more etails like how many messages got processed and message size and if possible may be deisgn change required to handle huge load like processing huge loads on interval basis not in one shot.