Setting up zimbra redundancy – live sync
It is important to have a failover of every system as it improves the availability of the system and reduces data loss. In this articale, we describe how you can have near-live synchronization between two Zimbra servers so that one of them is live and the other is kept in a warm or very warm standby state. The sync can work in reverse when the mirror (redundant) server becomes the active server. This allows easy fall-back to the original server once the failover condition is resolved.
Zimbra employs several different databases to store messages, message indexes, meta-data, account information and configuration. Although it is possible to synchronise two Zimbra servers at the disk level using DRBD or VSphere, the amount of disk operations from all these databases that need to be replicated would probably take up a lot of bandwidth which may be debilitating and/or expensive to implement if the two servers are in remote locations. Continue reading “Setting up zimbra redundancy – live sync”