Failover and disaster recovery scenarios are always important things to practice. If you've got a high availability service and you've got redundancy on redundancy to the point where it would take a global catastrophe to knock you out, that's all well and good. But if the last time you were forced to test that redundancy was 36 months ago, how do you know it's even still working?
We do backup verification, we do redundancy testing on a monthly/bi-monthly basis depending on the system, we failover our equipment on purpose to ensure that there is as little impact to the customer as possible when it happens because it will happen and we want to be ready for it.
But you'd be surprised how often I've walked into a data center and found that the key people involved in disaster recovery or failover scenarios had never actually been through one. Some actually had some spotty documentation that might help them cope in the event of a real failure, but many didn't even have that.
Testing the insurance just seems like common sense to me, but then again, maybe that's because I grew up on the gulf coast where preparing for disaster is a yearly practice that lasts for 5 months. Speaking of which, it's that time of year again (Hurricane Season).