Q. I have heard that using a private cloud can reduce virtual-machine sprawl. I don’t get it. If users are creating their own virtual machines, shouldn’t sprawl increase, not decrease?
A. In a virtual environment, there is a particular process for provisioning a new virtual machine (VM) for a business unit. The business unit puts in a request, and then an administrator manually creates that VM and gives the business unit the login information. This process can take days or weeks. Typically, there is no restriction on the creation of VMs, it just takes quite a long time. And in most environments, there is no tracking of which business unit owns which VMs.
This lack of tracking can lead to terrible VM sprawl. A year later, your IT department could be looking at hundreds of VMs, with no idea which ones are necessary or even which ones are being used—and no idea who asked for them.
When you implement a private cloud, however, the process is completely different. First, one or more clouds are created. Then, groups of users are given specific quotas on those clouds. The business units then create their own VMs within those controlled quotas. When a quota is hit, the business unit can ask for an increase. Or it can store existing VMs that are no longer being used in the library. Or it can just delete those VMs, to free up quota. In addition, your IT department can easily and clearly determine the owner of every VM in the cloud. This capability enables a concise scrubbing process, when required.
For these reasons, there is far less sprawl with a private cloud—and the total removal of VMs that have no clear owner.
And a private cloud can further simplify your environment by reducing resource silos. Instead of separate sets of compute and storage resources that are managed separately in a manual process, a private cloud solution uses a management stack that overlays all the resources in the data center. This management stack aggregates those resources into a logical view. Because services are provisioned within the aggregated resources, they can run on any resource in the environment.
Behind the scenes, you might still have separately connected groups of computer, network, and storage, but resources are correctly placed, based on the requirements of the service. Silos are abstracted through the management layer so that they no longer present a barrier to using the full capacity of your environment. And they don’t require special consideration when used.