While one of the most important benefits of server virtualization is increased consolidation, virtualized environments are not always as resource efficient as they should be. Even with the best VM sizing approaches and IT resource management practices, there will always be some resource waste. Periodic cleaning is crucial to avoid wasting storage and computer resources.
Here are two of the most common reasons virtualization cleanup may be necessary for your business.
Snapshots are very quick point images of a virtual machine on a datastore, allowing rollbacks to that point in time. Sometimes, VMware/Hyper-V snapshots can fail and if it is not corrected immediately, this orphaned snapshot may cause problems in the future.
To begin with, creating a large number of snapshots across virtual machines will consume a lot of disk space. In addition, the longer a snapshot exists, the more difficult and disruptive it is to close it. Second, if left unattended for long enough, the VMware/Hyper-V snapshot will use all available capacity on the VMFS data store, putting all VMs in that datastore offline – which is not good.
What Can You Do?
Keep in mind that if you aren’t using the snapshots, it’s a good idea to “delete” them after a set amount of time (Ex. every 7 days). Note that each time you take a new snapshot, the previous snapshots become read-only. Meanwhile, as modifications to the VM are made, the size of the snapshot file grows.
Snapshots should not be kept for any longer than your next backup run. Snapshots are no longer necessary once your entire system has been backed up. At that point you have a before and after backup that you can use instead of the snapshot. Remember – snapshots are best used for short-term rollbacks.
Create a policy for administrators when it comes to cleaning up old and un-needed snapshots. Continually re-evaluate this policy as the needs of your virtual machine change and evolve over time. You can also create alarms in vCenter that alert you when a snapshot grows past a set size, which is a quick and easy way of seeing where they have accidentally been left behind.
For many organizations, virtual machine sprawl, or VM sprawl, is a concern. The ease with which virtual machines can be created enhances the risk of a sprawl happening.
A VM sprawl happens when an administrator can no longer properly supervise and manage all of the virtual machines on a network. Administrators may wind up with too much to manage if VMs aren’t archived when they’re no longer needed. For example, this can commonly happen with file servers, and an organization ends up with 2, 3, or even 10 file servers all doing essentially the same job.
Furthermore, having these inactive VMs reduces network efficiency. These VMs consume resources on the VMware/Hyper-V hosts, both CPU, Memory, storage, and networking. Further they still need to be patched, backed-up and maintained or they can be security vulnerabilities.
Another form of VM sprawl is VM resources sprawl, where a VM gets more and more resources allocated to it over time that it may not need anymore. This can be fixed by doing a periodic audit of resources in use on VMs and whether they are actually using those resources.
What Can You Do?
Perform an audit of your VMs. This allows you to see which VMs are linked to a hypervisor cluster or a specific host. The goal is to ensure that you VMs are assigned to a hypervisor host or cluster. Look for VMs that have been removed from inventory but left on disk. It’s common to see 10-50% of the space in an environment being used by VMs that were removed from inventory but not cleaned up. Virtual machines that aren’t linked to a hypervisor can be archived or removed (subject to verification).
Begin VM archiving. Although deleting VMs frees up resources, it isn’t always the greatest solution. In some circumstances, a virtual machine may not be needed right now but will be in a month or two. In these cases, either backing the VM up to restore it later, or moving it to slower, more affordable storage is much more efficient. The most important requirement here is that the VM can be easily accessed again when needed.
Establish processes and policies for the IT team. Having these standards in place will make managing VMs and reducing sprawl much easier. The first step is to set roles and permissions for distinct users using role-based access management. These restrictions ensure that only authorized individuals can create VMs and snapshots, lowering the risk of rogue VMs. These procedures could also include requesting justification for any VM requests that are too large, resource allocation adjustment based on actual consumption, and adding rules for how many snapshots a specific VM should keep
There is a lot that goes into maintaining your virtual machines, and it’s easy to get overwhelmed. Following these basic steps and guidelines will help your business stay organized, secure, and efficient.