There was a joke that circulated years ago that if you wanted to get additional budget for IT you just said it was for ‘The Internet’. No one was quite clear as to what they would do, but they knew they wanted to be on, or in, or connected to it.
In the Pentagon that joke morphed. Whenever someone wanted to get additional budget, the reason was to ‘Deal With China’. Well, in fact, maybe that is still true.
In technology today, the current budget justification phrase is ‘Cloud Computing’. Except in this case, exactly what Cloud Computing is or what it can do is even less clear than normal. On the other hand, that lack of clarity means there are lots and lots of meetings, seminars, and conferences that deal with trying to define Cloud Computing and provide advice on what to do about it.
In that context, I was on a panel Monday, May 3, that discussed Cloud Computing and the kinds of new skills that would be needed to support Cloud initiatives, http://events.1105govinfo.com/Events/Cloud-Computing-Summit-2010/Sessions/Monday/CC4.aspx.
I had three major themes.
My first theme was that people tended to mean one of a number of radically different concepts under the general topic of Cloud Computing.
Many actually were talking about consolidating multiple applications on a fewer number of servers – virtualization. It was this step that accomplished much of the savings, if there were to be any, from Cloud Computing. In fact, it was certainly possible to do server consolidation and application virtualization without actually implementing anything that actually was ‘in the Cloud’.
Others used the term Cloud Computing to putting applications on the Internet; in the web. This approach is also often described as Service Oriented Architecture, SOA. I am probably not capturing all of the nuances of SOA but to me this basically means taking a program which traditionally was self-contained and isolated and treating it like a service which others could access or integrate into a larger set of combined services. Doing so efficiently requires writing programs a bit differently, adding the ability for a service to be discovered, that is found by others, and adding the capability to expose aspects of the service to others.
SOA in the end requires not just technology change but also cultural change. To be most effective it requires an organization to be much more collegial and standards based in how it designs and develops software.
Finally, some people meant having applications, or aspects of an application such as the platform it runs on, provided externally; that is, through a cloud. The big challenge here is that when using only internal resources it is possible, though in my opinion unwise, to get by without taking the time or applying the necessary rigor to develop service level agreements (SLA’s) for all of the aspects of your system.
You can tell if people are working hard by peering over their shoulders. You can measure performance by users calling and yelling at you, and dynamically reallocate resources by yelling at someone down the hall.
However, when you move a resource out of your internal operation it becomes absolutely critical to develop robust SLA’s to manage your provider’s performance and define your expectations. It turns out that this is very hard to do especially in areas that historically have not been defined in very precise terms such as security or privacy. This is, again in my opinion, one of the major underlying reasons why there is such resistance to moving applications to the cloud.
My second theme is derived from that last point. It was always useful to create business architecture’s to drive technology development. While it might be inefficient, it was historically possible when everything was accomplish internally to ignore that benefit and instead do what was in effect the opposite approach, develop technology solutions that ended up impacting the business.
However if an organization wants to move to the not-well-defined cloud, it becomes necessary to define the business architecture’s and business goals associated with the applications. Without that definition, the likelihood of achieving the promised benefits associated with Cloud Computing are highly unlikely to be achieved.
My third theme was that the major human capital impacts were:
- Technical and operational IT assets were likely over time to move to external service providers and away from user organizations
- The demands on procurement and legal professionals were going to change as their responsibilities became more and more ‘horizontal’ between organizations and their providers of service and less ‘vertical’ supporting internal hierarchical organizations
- The importance of technical staff who also were comfortable with business issues would dramatically increase.
Federal Computer Week, http://fcw.com/Articles/2010/05/04/cloud-computing-implications.aspx, covered the panel.