While at the Gartner Symposium in Barcelona, I had a stream of meetings with government clients from several European countries who are contemplating the development of private clouds.
In some cases, these clouds are seen as the end result of a consolidation process resulting from a decision to centralize IT infrastructure across multiple entities, while in other cases they are seen as an alternative to the current, often mainframe-based system to increase agility and reduce cost.
Let’s start with the latter first. In one case, an organizations that deals with periodic workload peaks assumed that going cloud would make their infrastructure cheaper. If it is a private cloud, and you are the only user, you still have to size your infrastructure on the peak requirements. The only way to use a smaller infrastructure is either to distribute those peaks better across the organization (which would be a business and ultimately regulatory decision in their case), or rely on external infrastructure (e.g. virtual private cloud) provided by an external supplier (either some government central entity or a commercial organization). Unfortunately in their case neither was an option, so the question remains: would they save money?
The former case – i.e. the relation between infrastructure consolidation and cloud – is also interesting. People are struggling with whether to create a private cloud by first colocating and then virtualizing existing servers, or by starting from a greenfield data center and gradually migrating workloads and decommissioning servers. However the real question behind is the following: should the resulting data center(s) be just a private cloud? Or should IT services be made available in a variety of ways, cloud being one (and gradually the most important), but not the only one? What if a department still needs a hosting service, or a co-location service?
So I can’t help observing that sometimes people self-inflict (or self’-prescribe?) cloud even when there is no clear need to.