HI Brandon
Mirroring, in my view is basically needed for a state-full fail over in your HA cluster. For it to be state full the standby device needs to know about the active connections on the active device.Also persistence mirror gives the flexibility of the when a session is no longer active for example user logged of the online clothing store with a cart full of items, when that same user log in again 2 minutes later (in which time my devices failed over) The now active ltm will send the user to the initial back end server he logged in before the 1st session closed, but you will have to consider the application it self. HTTP I am sure a quick 3way handshake will just continue a failover session when my active box fails, but a telnet or ssh session wont.
Mirroring configuration you will need depending your level of failover you need in your HA cluster, on the virtual server for the active connections, on the SNAT configuration for routing purposes I guess, and the actual persistence record as explained above. hope my input helps.