When I was in college, I was fascinated by a book called The Coming Crisis of Western Sociology. Actually, I was most fascinated by the title. Sociology was not my thing.
But now, finally, that thought has found a home. During the conversations I have with a wide variety of CITOs, it is clear that we are about to enter a realm of complexity that managers of IT at all levels are simply not ready for.
What is the coming crisis of IT management?
It is the fact that despite years of helping other parts of our companies implement software that provides a model of important parts of business operations, IT is relatively unmanaged.
Yes there are excellent tools for IT Service Management, such as Service-now.com, HP OpenView, BMC Patrol, and dozens of others. There are people that love or hate each one of these tools. But as we approach a new era in the data center, the tools aren't ready and we aren't ready.
First of all, most companies really don't know what is in their physical data centers. This may sound like an overstatement, but if you take a tour of a data center and point to a server and ask, "What would happen if we turned this off?" most data center staff would not be able to tell you.
To answer that question you would have to know a boatload of dependencies that are just not tracked. I had a fascinating conversation today with Jon Temple, CEO of nlyte Software, about the depths of this problem and how his software addresses the issue by discovering many dependencies.
But lots of dependencies are hard to discover. You would have to know what applications were running on the server, what applications depended on that application or services from that application, what network traffic would be affected, and so on. We don't know that most of the time and getting the answer takes a long time.
But that's just one aspect of complexity. Here's another one that is very hard to deal with:
When you create and configure server to run one, two or three applications or services, you have a variety of software that is installed on the server to allow the applications to run. If you are doing your job, scripts run at startup make sure that all the needed software is running so you can reboot the server and everything will work.
But what if you have to move one service or application from that server to another one? What supporting software should be turned off and uninstalled? What should be installed on the server that you are moving to?
While you might say, "so what?" to both of these questions, think about how we will need to run our data centers if we use private or public clouds. If the flexibility of the cloud is going to mean anything, we must be able to make changes faster than we do now.
There is a huge unanswered question about the cloud that I will leave for another time, which is, "What is the business value of the flexibility of the cloud?" So far, server consolidation, rapid provisioning, and new scalability models are what we are getting.
But how are business run differently because of cloud-powered flexibility? Examples are hard to come by.
My point is that as CITOs we should admit that we aren't ready for the new roles we will play and the new challenges the cloud and virtualization will bring. If the environment we manage becomes more virtualized, more dynamic, how will we avoid creating a mess?
If we put our own applications in the cloud or start using SaaS much more extensively, do we really know all the implications for disaster recovery?
Most of the time, no, but we are happy to trust our vendors, and most of the time there is no problem. But there will be problems.
I spoke with Rens Troost last week, a partner in Virtual Clarity, a London-based IT consultancy, and he set forth a new vision for IT as a broker of services. The CITO will have some on-premise resources, but will have a as a primary role the identification and management of services of all types.
The IT department will provide an environment with a variety of federated services for identity management and data, but there will be many services used from third parties. It is also likely that partners will share in the data center.
In this way, the IT department will play the role of the service aggregator that John Seely Brown described in the .
I submit that we are not yet ready to move to a more virtualized world or to use the cloud at scale, until we improve our game as managers of the resources under our control. Becoming brokers is a whole new game that we will have to figure out as we go along, but here again, we might as well admit we are not ready. I am not trying to be negative, just realistic.
What I hope to do at CITO Research, with the help of anyone wants to join in, is identify and help find ways of overcoming these problems.
I would love to hear from the CITO Research Tribe about other problems of managing complexity that will get in the way of IT becoming more flexible, powerful, and relevant to the business.
Comment on this post and let me know your thoughts.
-Dan Woods, CTO and Editor, CITO Research
Note: Evolved Media, sister company of CITO Research, has created content for Virtual Clarity.