[rt-users] RT Scalability

[Mark Chappell]

We use RT for a significant chunk of the admissions, enquiries and 
support queries.  We now have over 1/3 of a million tickets, 300 queues 
and 600 privileged users (not to mention a whole batch of users as the 
result of spam).

We've been running this off a dual Xeon box running MySQL (with a mysql 
slave backup server that can be swapped in at short notice), but as you 
might imagine, with that many users (a significant chunk of which use 
the web interface) it's been getting a little slow of late.

So we started looking at clustering options.

On the database side, MySQL cluster seems to insist on holding the 
entire database in memory, which with a 10 Gig and growing database just 
isn't feasible.  Oracle is just a little bit too much black magic and 
our DBAs seem a little reticent to play the clustering game.  So I 
started playing with PGCluster, and I now have a load balanced setup in 
the final stages of testing prior to deployment.

For the web side, Apache combined with Cisco switch SLB has worked a treat.

I'm pretty impressed with just how well RT has scaled, however, I'm 
curious as to how many other people have tried pushing RT this far.

Has anybody looked into the possibility of "archiving" parts of the RT 
database, leaving only a stub with no attachments content?


Mark
-- 
With sufficient thrust, pigs fly just fine. However, this is not
necessarily a good idea. It is hard to be sure where they are going to
land, and it could be dangerous sitting under them as they fly
overhead.
  -- RFC 1925