Effect of parallel processing on INDX query

[krishg]

We noticed that running costing run (SAPRCK10) with parallel processing dramatically increases the 'Elapsed' time for the DELETE INDX query (Navigation modules) in DB02. During a costing run in first 4 minutes, DELETE query on INDX table took almost 6000 seconds (most of it waiting). This query shows up as most time consuming query in DB02.

I was wondering if this is the cost of using parallel processing and there is no way to really optimize this?

[krishg · Accepted solution]

This was happening during the costing run. There is a OSS note that could potentially address this issue. This note requires customer enhancement. We will try implementing this note to see if it makes a difference.

1280119 User exit introduced to alter INDX key formed for Octroi