Open any IT magazine today and there will be at least one article, and five advertisements for some new cloud computing product or service. It is not even fun anymore to talk about how big The Cloud will one day be, now how it might change the world. The only interesting things left to think about are what might the uber cloud look like, what will the various flavors of clouds that create it be (public, private, specialized, milspec, etc), and how should organizations migrate to it. Most companies are already headed there because they are using at least one SaaS application, so even if the executive team wouldn’t know a cloud from a access control list, they have one foot outside the tent.

This is causing me to think about what the best practices might be for migrating an entire IT department to the cloud. Set aside fears about controlling data, security, bandwidth, and the IT people standing in the way of a transition for a second and let’s just assume that cloud based computing gets 100% dependable. Then take an even bigger leap and assume that your IT people would actually disintermediate their own jobs. If all that could come true, you are then free to dream about what the best practices for the transition might look like. How fast might it be done? What systems go first, and which go last? How do you measure the ROI? How might you modify contracts with cloud providers to protect yourself?

I have been working on a long process document that would be used to dictate an orderly step-by-step, and find that there are really pretty simple answers to these questions. The first hurdle to really get over is just getting your hands around the fact that cloud providers can actually do IT services better than you in most cases, and that you can dual home your bandwidth so that you do not have to worry about access to your data and applications being cut off by the random cable cut. Yes, your data will be secure, and in fact, the cloud providers IT people are better at security than yours and probably have more to lose. Yes, it is possible for the applications to run just as fast over the wire as over your internal network, and the added bonus is they will run great anywhere in the world instead of forcing people to VPN to the network. Let’s be honest, having to VPN into a network is a bit like trying to get an appointment to see your doctor – sometimes you get in, and sometimes you don’t.

Once you get past the mental blocks and years of hosting your own servers, you can think rationally about what to do first, what to do last, and how fast to move your whole software infrastructure. The steps could really be reduced to the following: Study your usage patterns and identify all software applications in the organization. Rank then in order of their unique importance to the organization. Then make a decision as to whether you can replace the application with an existing SaaS option, or whether you will just need to move the current application into the cloud. Find a general cloud vendor that might fit your profile for services, and calculate the costs of renting their services. Examine the costs of the SaaS systems you might be able to trade over to, and the general cloud services where you cannot go SaaS. You now have a projected budget number you can play with to see what kind of ROI you might get (if you also are able to lower your staff count.) If you like what you see, you can then form a plan as to which applications go first and I would suggest start with the smallest and easiest and work up to the hard ones. Then measure performance differences after the moves in order to gauge if you have taken a step forward or backward. Oh, of course it is a bit harder than this, but I only have a page or so to make the point. And that is that every organization needs to have a solid transition plan and some goals in place. Also, that it is not that hard to think through. The problem in most cases is that the business side of the house does not know enough about the cloud to demand a move, and the IT people have a vested interest in keeping things in house in most cases so the only time you see an aggressive plan to move is when the budget gets slashed and there is no other choice. It is just sad if it always takes a financial tsunami before organizations figure out a cloud plan.

Scott Klososky
Scott@klososky.com