What are the hidden costs when migrating from on-premise to cloud computing?

Bill, what do you think about costs associated with business changes that result from the change to being an "on-demand organization." Or, are the opportunities for innovation so strong that the costs and risks are not reallly relevant?

You have a point in that there is relevancy for some organizations but there is no general rule. Information Technology is, in its simplest form, process automation. There are 3 ways organizations address this:
1. Develop automation around an existing and proven manual processes;
2. Buy software and compromise some of your processes to use the software; and
3. Buy software and change your processes to match the application
There’s actually a fourth one in which you buy software and modify it so it looks and operates like your organization did in 1965 but that is not generally found in the commercial world.
Moving to the cloud is simply having the “machinery” somewhere else and has no real bearing on process. Now if you’re talking about a new application or SaaS, there is a cost associated with it. But the cost to change the process (what people and machines do) is the same whether it’s in the cloud or in your closet.
Finally, there is the potential of changing the “personality” of your organization to be able to quickly adapt to market changes. This happens over time – not because you copied software and data from a hard disk in your office to a hard disk in Iowa – but because you have developed a change process; proven it out by making several changes; and improved the speed of change. The cloud allows you to do this by removing the limitation of IT investment amortization; or keeping the old car until it’s paid for.
Change is a two-way street in that you can change for the better or, errantly, change for the worse. I remember someone telling me that anything that you can change in 24 hours can be changed back the way it was in the same amount of time. So your executive team really needs to be on the ball and out in the market constantly to ensure that the change decisions have some basis for them. But each change either has an associated cost (and benefit) or the cost is irrelevant like moving from a U.S. manufactured light bulb to CFL bulbs only made in China. The cost is irrelevant because Congress mandated it with a deadline of 2014 and the last manufacturer of incandescent bulbs closed in 2010.

Bill, I completely agree with your thoughtful points! It's certainly true that taking advantage of changed ways of working requires a substantial culture change over time.

With respect specifically to the cloud, do you think there are benefits and process changes that require those kind of long-term culture shifts to which you alluded above?

Excellent observation!

Not at all – I don’t see any process change! Information Technology, in its simplest form, is the automation of business processes. Cloud computing is “where” the machinery is and has little/nothing to do with business process (typically applications). Change in the IT department can occur if you move to a model in which you pay your vendor to manage the operating systems, security, storage, et al. In that case, you need fewer people or maybe different skill sets.

The first benefit of cloud is that you no longer have to procure and maintain the hard assets. The second benefit might be that you pay for usage unless you require a dedicated VM. In most of my experience in the commercial sector (I’ve also been in the Federal side), the business units never really had to wait for the hardware and O/S. IT was pretty much in place at, or before, the application was ready.

Culture shifts will happen when the business units recognize 4 things:
1. The market is changing;
2. The cloud eliminates a capital spend to react to the
market change;
3. The business unit can now react faster than
before;and
4. Someone in the business really understands process
automation (Should be the domain of IT).

One of the concerns I have is that the same problems keep cropping up but they are called by different names. Remember “server sprawl” which is the proliferation of servers throughout the data center? Now we have “VM sprawl” which is making things more difficult because you can’t point to a box and say “This is the payroll system”. Today, you can’t really find it. The next problem is going to be “cloud sprawl” which will make the former 2 problems pale in comparison as you won’t know where on the planet your servers are and the associated business processes they run. Heck, you won’t even be able to inventory your applications if each business unit goes out and contracts for their own cloud service. In each case, the solution has been to consolidate but in cloud sprawl I don’t think you will be able to find all of the servers in use.