I had my first exposure to VMWare in about 2000, when it was still a free desktop product, and if I remember correctly, the version number was below 1.0. Since then, I have watched it mature through several stages of acceptance. I saw my customers pronounce it fit for test / dev, (but not production), then certify it for disaster recovery, (but still not production), and eventually second tier production. Today I hear of people running their entire operation on virtualized servers. This trend is familiar to me from the Linux and even the PC adoption curve.

Today we see the same cycle repeating itself with Cloud Computing, especially Infrastructure as a Service solutions. Some say this approach is only good for test / dev, or second tier applications. I say it is inevitable that significant chunks of the world’s IT workload will happen in the cloud. Ten years ago, we were having this conversation about where to host the corporate web server. Today, almost nobody keeps their public web server in-house, because the quality, reliability, security, and cost is so much better from the specialized providers.

So the move to the cloud is inevitable, but that doesn’t mean you have to abandon your data center today. There are still kinks to work out, and those short-term kinks might be very good reasons why you cannnot, and should not move all your IT operations whole-hog into the cloud. But your DR operations seem like a slam-dunk for the cloud.

I have been in the DR business for over 11 years (not counting my years as an IT manager, struggling with, and typically failing at DR), and I am the first to admit that DR is a pain. The first half of this post argues that you will want to give your data center over to specialists, but disaster recovery planning tells you to go build a second data center! The second thing your DR plan calls for is a bunch of servers that you aren’t going to use, except in a disaster. And data centers full of servers require planning, investment, build-out, on-going operating costs, and then updates as equipment ages. All of that can be rented from the cloud, on a pay-per-use basis.

In the best case scenario, you can get access to the necessary infrastructure for no up-front investment, with little or no configuration / build-out time, but have access to nearly unlimited resources when the need arises. Best of all, you only pay for the capacity you actually use – you don’t generally pay for any of the stand-by capacity. Cloud computing can be a confusing jumble of too-good-to-be-true claims, but for disaster recovery, it looks like a perfect fit.