Performance Degradation of new Servers

[Former Member]

Hi All,

We are experiencing massive performance degradation on production when the system is under heavy use. It seems to be at its worst around month end. The worst effected transactions are the Cost / Profit centre 'line item' reports using RCOPCA02 (and other similar programs).

We have the message server running on the database instance as well as 2x App servers. The Msg Srv and One of the App servers are very similar builds

4x AMD Opteron 875

16gb RAM (10gb Pagefile)

Both running Win Server 2003

We're using MS SQL 2000

The other message server is a bit weaker but has been around for some time (few years) and hasn't caused any issues.

We have recently moved the Msg Server from an old (much weaker) server to the new build and since then seem to have performance issues. Initially after the move we had issues with the number of Page Table Entries (where down at about 6000-8000). Using the /3GB /USERVA=2900 switches we have this up to about 49,000.

If anyone has had a similar experience or could offer some assistance it would be much appreciated!!!

Cheers,

Kye

[Former Member]

While investigating a different issue we've found that we have 3 servers that share the same set of disks on the SAN. Two of these where Message Servers (R/3 and BW) when both of these where running the 'Average disk queue Length' was at times reaching 400! (should be around 1-3).

We've moved one of these instances back to DR which has relieved some of the pressure from the disks.

We have also added another index to the GLPCA table (which was causing most of the problems).