Teamsite performance

felipel

I'm looking for suggestions on how I can increase Teamsite performance. I will describe our current configuration in as much detail as possible, and if anyone has suggestions on how to improve performance, they would be greatly appreciated.

My company runs Teamsite 5.0.2 (not SP1) on a 4-processor (P4) Win2K server with 4GB RAM. We have approximately 300 users with anywhere from 10-30 simultaneously using TS. We have 8 branches, with one branch being our main Intranet site. That branch has approx. 30 workareas. The branch has approx. 11,000 generated/templated files in Staging (and that number is expanding rapidly as we convert old content into TS). This branch has approx. 500 editions with another 5 editions created each day. The other branches are much smaller, with approximately 1000 files in each.

As the size of the branch has expanded, we have experienced pretty significant performance degredation. At times, simple operations like navigating the file structure can take 10-30 seconds. More complex operations like creating/saving a DCR can take 1-3 minutes. Global operations (List Modified or List Locks) take an extremely long time, sometimes timing out.

In the scenarios I've described above, we had the performance settings set at:
cachesize=60000
rpc_threadcount=64
fs_threadcount=4
fs_active_area_cache=8

After doing some reading, yesterday we modified the settings to:
cachesize=250000
rpc_threadcount=64
fs_threadcount=5
fs_active_area_cache=20

This seems to have helped, but I'm wondering how if there's more I can do. We plan to upgrade to 5.5.2 at the end of November, which we've been told will help immensely. Any other suggestions in the meantime?

2 other related questions:
1. We've noticed that when the server really drags, it puts "[BUSY]" next to many of the workareas. I haven't been able to find any documentation on what this means -- anyone know?

2. Is there a way to disable the failsafe setting? Or possibly make the number of operations it stores much larger? We're not concerned with maintaining transactions in a failure, especially if it's a large amount of overhead.

Thanks in advance for any suggestions.

felipel

Just an FYI update to my own post in case anyone else has similar issues. I finally found buried deep in the Interwoven documentation a reference to a limitation of 500 editions on a branch (also applied to 500 files in a directory, but we were hitting the editions limit). We wrote a custom script to walk the branch and rebuild it while maintaining work in process, and that fixed the problems. Looks like this will be a regular process until we get upgraded to 5.5.2...