Best pattern I've seen to guarantee backup availability is for upgrades/deployments to occur on the hot failover first, swap the hot failover to primary, move p

Best pattern I've seen to guarantee backup availability is for upgrades/deployments to occur on the hot failover first, swap the hot failover to primary, move primary to hot failover status, then upgrade what was previously the primary. This, combined with your general data backup (database dump to tape, etc.) gives you much more resilience should something go badly. However, it doesn't take the place of regularly testing your data backups, and only works if you have a full failover environment.