I spent a frustrating day last week waiting for someone to do something to bring a server in an unknown location back online. The system was offline for 20 hours and my one contact was unable to explain what had happened.
Well, Steve, what have we learned from this?
Now that the system is back online, you can be sure that I'm backing it up to a machine I can touch and see, and I'm building a clone on a VPS at a company that's been immediately responsive to every tech support query I've made and gives me access to the entire virtual machine. It really comes down to trust and verification. And complacency. If you know something is wrong, fix it. Just because the bad thing hasn't happened yet doesn't mean it won't happen a little later today.
A service level agreement is only as good as the counter-party's ability to execute. If you have no faith in that, there's no SLA, no matter what might be on paper.