A few months back, we noticed that we were taking hundreds of RID pool sort failures on commonly executed retrieve queries going against application group tables. Given the nature of how the indexes on these things are established from IBM (many one column indexes), we came up with a tweaking of one of the indexes in order to make it a multi-column clustered index (POSTING_DATE, OD_RID_TITLE, first column of the application group table) so as to eliminate/reduce the hundreds of RID pool sort failures we've been seeing. Doing production tonight, so will see how successful it turns out.
Just curious if anyone else has had similar experiences (btw, we are on version 7.1.2.9)? I've heard that version 8 of OnDemand has the ability for the client to create multi-column indexes when creating new application group tables? Assume with the CLUSTER keyword as well?
Appreciate any insight. thanks
