> So, instead of 2 DBMail boxes, I will have one Cyrus box (which the users will connect to), and a DBMail box which will only be used for statistical analysis and archive purposes. DBMail is perfectly > suited for that purpose; it saves me a lot of coding.
Unless you've got a particular need to run Cyrus, why not try the following: 1) use DBMail for your primary server which your users log into 2) set up one-way replication from that database server to another box 3) Direct all search queries/etc to your slave database, which means no load is incurred on your primary box We're just in the process of setting up something similar, where we scan through customer mailboxes for say people over quota limit, or inactive mailboxes, etc - and can't afford to have the load on the primary server. Then when we find a message, mailbox, user, etc or whatever we want to delete, we can direct the update query to the primary box. This moves all the searching load to the slave, and only puts minor update queries to the primary server. The bummer at the moment, (according to list popular consensus a few months back) is that you can't use either MySQL server as your 'primary' database, as the replication is not real-time enough, so you may end up with out-of-sync views of data. But treating the slave as read-only, should negate/solve this. -- Regards, Mark Mackay
