Paralllel processing for multiple users in parallel

[Former Member]

Hello All,

I have a scenario where users on executing a particular FM will get the records from database(FM is user specific and fetches the records for that user only), since the FM used to take long time because of teh number of records in backend, we have implemented paralllel processing for the FM. Now for the user the parallel processing is working fine and based on the calculation we are doing internally the system is optimally using the free workprocess.

The issue here is, the parallel processing is working fine for single user but as soon multiple users start executing the same FM as FM is user specific the system tries to get all teh resource available in the system and once all teh work processes are allocated it gives RESOURCE_FAILURE error.

For example if in the system I have 20 work processes and for the first user if it consumes 4 work processes and at the same time if another user executes the same FM for his user then the system will allocate another 4 work processes and as soon the number of users increases the workprocess utilization will increase and results in RESOURCE_FAILURE error.

Kindly let me know in case if any one of you have encountered similar type of issue, it would be great if you could provide a solution for this issue.

Thanks and Best Regards,

Neerup.

[Arun_Prabhu_K]

Hi Neerup. I have not faced an issue like yours but here are my few ideas.

1) Make username in the table as a secondary index to the DB table.

    Data fetch will be faster.

OR

2) Create a lock for the DB object when one user uses accesses the FM.

[Former Member]

Hi Arun,

Thanks a lot for your suggestion, but that is not going to work for me, the thing is the FM will do lots of processing and calls to different FM and is not a simple db call to fetch the record, I told the same initially so that its easy to understand.

Kindly let me know if there are any more suggestions, will be happy to receive.

Best Regards,

Neerup.

[Former Member]

Hi Neerup,

It sounds like your servers are lacking in either RAM or processes - one option would be to boost the system resources.

If you are running these functions in batch jobs, you could try routing them all onto a single application server and limiting the number of batch processes on the server - this will cause them to queue up, stopping too many running at once and using up all of the available resources.  You can specify the target server as a parameter in function "JOB_CLOSE".

Regards

Colin