We have seen significant interest in monitoring VMware recently. Server virtualization has gained momentum for use in physical data centers, as well as, in the cloud. Where it is deployed, it has had a positive effect on costs.
The proliferation has caused us to consider some of the challenges that beset VM deployment:
- Not enough expertise
- VM sprawl
- Capacity planning
- Licensing compliance
Many organizations lack expertise with virtualization. While large IT operations may have people with the required experience, smaller organizations may have to contract for the expertise. In this case it becomes important to manage the knowledge transfer to in-house resources.
After moving down the path of virtualization, how can VM sprawl be controlled? With the virtualization expertise, it becomes easy to virtualize machines and applications. Even when virtualization doesn’t make operational sense, or where it is unnecessary, it can grow out of control. In the extreme case, just because you have a site license for software, it doesn’t mean that each person should have their own virtual image.
Uncontrolled proliferation of different virtualization schemes – VMware, Microsoft, Oracle, RedHat, EMC (Citrix), Amazon – is yet another type of VM Sprawl. Each base requires another area of expertise, and each base introduces the potential for issues for later moves or migrations.
VM sprawl can lead to confusion with capacity planning. If images proliferate for convenience rather than necessity they can incur additional capacity without a justifiable productivity gain. Also in this case, it is possible that dormant images could further impact capacity.
Virtualization also brings challenges with license compliance. Procedures need to be tight to know what images are checked out and where are they running. If an organization has a single seat license and an image of the software is unknowingly dormant on one virtual machine, does another image of the software instantiated on another virtual machine create a license compliance issue?
In summary, deploying virtualization requires training, a process to know what images are running where and a policy restricting the use of virtualization to cases where it is operationally required or justified.