Sunday, 20 March 2011
Where there are Clouds there is Rain
We have probably all heard the wonders of Cloud computing from pre-sales teams. Cloud computing will reduce energy consumption, reduce costs, improve innovation, and all kinds of other things.
Well over the past two years I have seen a number of Cloud projects at various orgnaizations, and none have worked out as intended.
My impressionis that the Cloud is being way way oversold. Why? Because even though the Cloud makes sense to IT providers and outsourcing firms as a way to potentially increase profit, there are no real benefits to most clients. In fact working with the cloud, especially the current Microsoft Stack in the Cloud produces complex issues around security, access control, sign off, and boundaries of responsibility that significantly slow down project deployments. Clients generally find themselves waiting much long for a Cloud solution than they would have a simple data center.
This is not to say that the Cloud is dead, it certainly is not. But I have restrained my expectations for what the Cloud can offer a great deal in the last year. The key problem is security. Generally the requirements for security in a Cloud solution handcuff the solution and make it almost impossible to move forward.
The Cloud has read head first in to IT's dirty little secret. In data centers we generally put the software in and spend the weekend before fixing all the mistakes we made. If we have direct access to the boxes we might be able to meet our deadlines. IT projects are still riddled with mistakes, oversights, and miscommunication between parties and players. That is simply the nature of complex projects in technologies changing all the time. Having a single data center helps to resolve many of these issues at the last minute.
But the Cloud often isolates the production environment from the people who can make it work. Software is deployed in to solutions beyond the deployment teams control. The inability to "pound it with a hammer" at the last minute has bogged a number of projects in endless efforts to document everything before hand. I am seeing very basic projects now producing hundreds of pages of documentation which consumes months of work.
So far I have yet to see a Cloud project that did not take at least 5 times as long as a traditional data center project. My hunch is that unless a set of Out of the Box Cloud services are soon developed and sold at lower cost than software licenses, cost conscious consumers will simply return to what worked in the past: installing software on servers. As server become cheaper, faster and more powerful and as SharePoint becomes more and more the standard office management system it will actually become easier to keep IT in house. And what I am seeing now is that there is a strong movement in that direction.
I am not prepared to say the Cloud has failed, only failed to be the miracle it was promising.