Disk Space Usage: A Hidden Cost of Virtualization?

Wes Noonan (Profile)
Tuesday, April 5th 2011

A recent study by Enterprise Strategy Group, entitled “2011 IT Spending Intentions Survey,” noted that more than 60% of surveyed midmarket and enterprise organizations plan to increase their virtualization software spending in 2011. While virtualization is largely expected to help reduce operational costs, there are two particular hidden costs that often present administrators with an ongoing challenge: disk space usage and poor disk performance.

The Hidden Impact of Snapshots

In many IT organizations, server administrators are allowed to create snapshots of their virtual machines (VMs), which enables rollback of modifications if they go awry. However, snapshots can consume significant disk space, the extent of which remains hidden to most users. To make matters worse, each time a user creates a snapshot, that snapshot begins collecting the changes made to the server since the snapshot was originally created. It is, in effect, the delta between “now” and when the snapshot was taken.

If a system has multiple snapshots, each file must be opened and processed to recreate the I/O data for the virtual machine. This only exacerbates this scenario and means that every read or write operation is multiplied by the number of snapshots, creating contention and poor performance of the underlying disk subsystem. In addition to disk space considerations, administrators have to deal with performance degradation that can impact the entire host and all VMs running on the host/datastore.

While a user may be constrained to a 10 or 20 GB disk when creating his or her VM, snapshots are not similarly constrained and can grow many times larger than the original disk. Unless the administrator works with each user to either commit or revert their snapshots, the virtual infrastructure can quickly run out of storage, requiring organizations to purchase more disk space to accommodate these growing snapshots. Additionally, the performance degradation that can result from snapshots can oftentimes be masked, causing organizations to misattribute their issues and potentially invest in faster and more expensive storage than is required.

Deleting a VM for Extra Disk Space? Not Exactly.

Deleting VMs does not always result in a complete return of the disk space that a VM was consuming. Users have the ability to delete a VM’s configuration from the VMware User Interface without actually deleting the files stored on the virtual infrastructure – allowing users to recover the VM at a later time if needed. However, this leaves orphaned disks resident on the virtual servers and without a deep investigation (often taking hours of effort), virtual administrators often have no idea these orphans exist and, as a result, they can consume untold amounts of disk space.

Automation = No More Hidden Disk Space and Utilization

IT process automation is a proven means of controlling the costs of hidden disk space and utilization across virtual environments.

By automating the snapshot request process, administrators can establish policies for the length of time a snapshot is allowed to remain in a particular place. Users can be given the option, through a self-service portal, to revert the snapshot, commit the snapshot or request an extension – all without requiring the virtual administrator to keep track of individual requests.