Ironport clusters not like traditional clusters

[lfkentwell]

I know the Ironport cluster is not like your traditional cluster where you have a VIP and redundant hardware to ensure seamless failover should a device fail.  Im keen to hear then how others have achieved a Highly available state for incoming emails to 2 or more Ironport appliances with seamless redundancy.                  

[Tim Boswell]

For inbound mail, MX preference is fine for most purposes.

For outbound mail, you can either take the same approach (smarthost preference) if your groupware supports it, or use a third party load balancer to provide a VIP.

[Donald Nash]

We use a load balancer in front of our ESAs. Look in the user guide (the advanced one, I think), for "direct server return" load balancing.

Using a load balancer makes for practically instantaneous fail-over. Granted, thats not as big a deal for SMTP as it is for other protocols, but it does avoid unnecessary mail delays. It also allows new ESAs to be deployed without DNS shenanigans, since the MX record refers to the load balancer. That's very helpful when you're protecting several hundred domains like we are. Updating hundreds of MX records is a drag.

++Don