Eric M Russell (4/17/2015)
I was thinking along the lines of intermittent failure, something that causes a disruption of a few hours or maybe a full day (which might have more to do with our ISP than the cloud data center), in which case we owe the client an apology and a few thousand dollars.Even with a local instance of SQL Server 2014 on standby, it can take hours to restore from backup and more than a day to get the applications operational. The whole idea behind leveraging a cloud data provider is that the organization doesn't have to invest in the hardware, licenses, and staff required to host the database locally. For many, perhaps most, falling back to local hosting may look good on paper and help management rest easier, but in reality it would only mitigate a scenario where the cloud is unavailable for a week or more, and it would no doubt involve a huge financial cost. It would be equivalent to the aftermath of corporate headquarters getting flooded.
I claim no expertise, but I recall hearing about load balancing/failover hardware that if one cloud provider failed, it could switch over to another. The problem is that no two cloud providers work the same (that may no longer be true), so you have a lot of setup logistics to keep the two clouds in sync.
Most businesses can survive the inconvenience of hours or a day of being disconnected from their app cloud, though there's going to be lots of mutinous rumbling. I think if it's a problem, then your business model may be flawed and you probably need an IT department and staff. At least if your HQ is flooded, it's a tangible reality that people can envision and participate in fixing it. If the cloud is down, you're screwed, there's nothing you can do, and good luck getting incident updates that are translatable to managementspeak.
-----
[font="Arial"]Knowledge is of two kinds. We know a subject ourselves or we know where we can find information upon it. --Samuel Johnson[/font]