Yes, but the requests aren't the problem. State sessions, concurrent index writes across slow shared file systems, in-memory caches and lots of other things come into play. That's why Data Centre was written - simply slapping Atlassian applications on to allegedly clustered solutions either caused data corruption or was unbearably slow (usually, both)
There was a good write-up for non-Data-Centre JIRA/Confluence somewhere, but the bookmark I've got for it is restricted now, and the searches take you to Data-Centre stuff now, which is rather annoying.
However, it boils down to:
1. Put your attachments on replicated storage (never the index though!)
2. Have a second server with an identical JIRA installed, but not running
3. Use database replication so that your failover rig has a realtime copy of production data
When things go bang, you can
1. Shutdown production JIRA, break the database replication
2. Point DNS to the failover (actually, your load balancer could kick in here, but don't quote me)
3. Start the failover system
4. Reindex it
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.