The Cloud Computing freight train roars on, if not with a lot of motion, at least with a lot of noise and tumult.
I have been part of a number of panels over the last few months which focused on the subject of Cloud Computing, the current state of the’art’, and as usual what barriers exist that need to be dealt with to make it easier to utilize. There certainly has been much written about it both pro and con and it remains a high-priority focus for the current Administrator and, in particular, Vivek Kundra, the Federal CIO.
For one of the graduate classes I am teaching this semester at the University of Maryland University College, the subject is touched upon as part of a broad look at technology changes and implications. The topic generated much comment by my students.
It seems to me that the subject starts from the wrong side of the discussion, the technology side. When the discussion turns to the impact, it starts at an important but not the most important concern, that of return-on-investment (ROI).
Today and in a number of future blog entries, I will talk about what I think are the current important issues associated with cloud computing. Today I start with what I perceive as a foundational issue, cost, but later in the week will move to what I believe are more important considerations and goals.
Note: I do not plan to rehash what cloud computing is, or is not, there are too many other write-ups that do this. Look at the National Institutes of Standards work on such definitions, I think it is pretty good.
SAVING MONEY. To me the least important, though I hasten to say not unimportant, goal of cloud computing is to reduce costs.
In the simplest sense, the provisioning of IT services costs money because of an overhead cost associated with buying computers and putting them somewhere as well as the operating costs of running them. When you spread that cost over more users then the cost per application usage goes down.
With cloud computing you have the potential, emphasis on the word potential, to achieve these savings by running multiple applications on the same computing equipment. This can be achieved when you use techniques to allow more than one application to run at the same time on the same computer increasing its utilization or when the peak levels of demand are different for each application, or both.
This much can be achieved by using what is called a private cloud, that is one that you run yourself. For organizations that have not centralized the provisioning of IT services, this one change can have a significant cost savings. The barriers to doing this are to some extent technical, it is necessary to gain experience in how to do this; but in large part cultural and organizational, it requires different groups within an organization to plan and work together.
Historically, computer usage in data centers is amazingly low, on average between 5 and 15 percent of capacity. By running multiple applications at once, using techniques such as virtualization, this capacity usage can usually be brought up to over 50% and often higher. This reduces the need for additional computing resources and cuts down on environmental costs such as cooling and power.
Moving to a more public cloud, which is one provided outside the organization, has the potential to achieve greater cost savings (maybe). Again looking at this in the simplest fashion, it spreads the overhead cost across still more users, with a public cloud perhaps in the thousands or more.
The other added advantage is that those organizations who have recognized that running data centers is not actually their core competency can out-source, currently a politically complex word, data center operations. On the other hand, organizations that do so need to develop a core competency of working with outside providers, which many organizations do not do.
It is interesting also to realize how many organizations want to move to externally provided IT resources because they feel they are too disorganized internally. This hope generally is not realized. There is an old saying that IT cannot organize a disorganized situation. I can promise you that outsourcing IT will not bring management controls to a situation where none currently exist. You first have to organize internally and only then look for outside provisioning. Electric power providers do not untangle the wiring in your house.
This last step, moving from an internally provided centralized IT provisioning process, private cloud, to an externally provisioned process, public cloud, is made still more complicated for Government due to security and privacy issues as well as fear of embarrassment issues; who needs to read in the Washington Post that your personnel system was hacked while sitting on some public provider; I speak from personal experience that testifying on the Hill about security issues is not why most people go into public service.
Having said that the first step, centralized provisioning, achieves a large percentage of the gain, and is worth working toward.
The other challenge associated with saving money is that many organizations do not do such a great job of tracking the costs which they are trying to reduce. Government organizations in particular often have in place rudimentary, or non-existent, cost accounting systems which keep track of all of the overhead associated with doing such work in-house. Thus the cost comparisons relate lower than actual internal costs against actual external costs. It is not a surprise that different organizations produce different conclusions.
I used to joke when I was at the US Department of Transportation that if you wanted to achieve a certain ROI I could help do so with 10 minutes and Excel.
To be continued …