I have recovered using backups... Not actually due to a hardware failure .. but moving around servers. We took the backups, then reapplied the backups to new server. I HATE proprietary backups from the likes of Symantec - but that is my own bias... and they are probably fine. I have used (using) FBackup .. worked just fine. We have backups every 20 minutes ... and "full directory image" backups every 3 hours.
Directory structure is everything... AND simple stupid QUICK recovery is a must. We already have scripts built and available to recover quickly.
I totally AGREE that one has to actually perform/use backups and then recover from them. Its better to do it and practice - then discover issues while under stress.
Couple of other points:
#1. Because SM does the "force uninstall /install" for upgrades, . for our email servers... we could care less about total system recovery.
#2 Only other issue that has "bit" us and bit us hard is: IIS settings. Since webmail is a big critical part.. the recovery and implementation of IIS setting is a whole another story. Having learned from the school of hard-knocks; we do the following: Using IIS export feature ... we export via script every 24 hours - the setting of the IIS machine to our "stand by machine". As part of recovery -
a.) Apply data/folder structure
b.) Upgrade/Install - SM - pointing to current directory structure.
c.) Apply IIS settings and certificates.
d.) Start validation :)
So to sum up our "disaster recovery"
1.) We have a cold machine on stand-by.
2.) We take incremental backups every 20; full every 3 (am sure there are different and better
recipes) - this is just us.
3.) Export IIS settings (for all the domains/settings related to SM Webmail) every night.
in the event or in rotation for disaster recovery process.
a.) Start cold machine.
b.) copy over current image/then incremental(this already a batch script written).
c.) Unistall/and /or Install - same version of SM.
d.) Import IIS settings
e.) Start server.
On our best day... think we completed this in about 35 minutes.
To that point - we have external gateways for inbound emails