Virtualization’s claim to fame is the amount of space it saves in the data center and the efficiencies it creates by eliminating physical servers, conserving energy and even contributing to lower impacts on global warming. Virtualization can also be a data center manager’s best friend at budget time because it is easier to justify investments in virtualization technology when you can immediately show return on investment (ROI) with easy-to-see impacts on data center floor space, and with energy bill reductions of between 50 and 70 percent.
But virtualization also creates inefficiencies that sites tend to overlook. Why? Because there is a tendency for data center managers to think that they are home free once they have made the transformation to virtualization and they have realized the initial impact of data center savings.
This immediate sense of satisfaction creates complacency that encourages sites to fall into bad virtualization habits. “We see this more often than we’d like to,” acknowledged a major virtualization vendor, “in fact, we have some major engagements now with customers where we are helping them to combat virtualization sprawl.”
How does virtualization sprawl happen?
First, the ease of resource provisioning in the virtual environment makes it almost a mindless exercise for IT (or even end users in some companies) to allocate a data center resource like an operating system in the virtual environment in minutes. This ease of provisioning virtual resources makes it more likely for data centers to end up with more virtual resource allocations than they are actually actively using.
Second, most sites are focused on time to market with their virtual system deployments. This is what they have set their SLAs (service levels agreements) to and what they are measuring. No one is watching the backend of this process—or asking questions like, what do we do with a virtual resource after a user finishes all of its work with that resource?
The end result is the virtual sprawl in data centers—a phenomenon that begins impacting more than half of virtual server implementations within the first year of virtualizations.
Because of this, getting on top of virtual sprawl before it becomes a problem is an important strategy that every company should include in its data center best practices. Sites that fail to do this start losing on the all of the gains they initially achieved when they first virtualized.
How can sites effectively prevent virtual sprawl?
To ensure that virtual sprawl doesn’t occur in the first place, meet with end users so everyone can agree to a set of guidelines for deallocating virtual resources if they go unused for a certain length of time (e.g., 30-day non-use means the resource gets deallocated). The workflow for this should be organized in such a way that there are no surprises. This can be done by having a system that automatically notifies the end use of an impending resource deallocation several days in advance—giving the end user the option of notifying IT in advance if the resource needs to be retained.
In the case of virtual sprawl that is already there, there are no immediate shortcuts. IT must sit down with end user areas that these resources are allocated to and gain agreements on which resources are to be deallocated. The goal of these meetings should be mutual agreement on virtual resource deallocation guidelines that work for both IT and the end business.