How to Reduce Your VMware Disk Storage Use

Written by Joe Kozlowicz on Wednesday, June 25th 2014 — Categories: Cloud Hosting, Cloud Hosting, Cloud Storage, VMware, FAQs

As you grow more comfortable in the cloud, you’ll likely find more and more of your applications making the transition to your VMware environment even as your existing virtual apps continue to grow. Add backups or DR to the list and your storage use might start to get a little out of control, especially with linked clone virtual desktops or snapshot trees that save multiple VM states.

Storage costs can add up, so you’ll want to stay proactive in maximizing disk usage and eliminating inefficiencies. There are a few ways to maximize storage and reduce disk sizes depending on the VMware product and deployment:

vSphere VM Provisioning: Use Thin Provisioned Disks

There are three types of virtual machine disks (VMDKs) to choose from when provisioning a new VM: thin, thick, and EagerZeroedThick. Unlike the latter two options, thin provisioned disks do not reserve space on the Virtual Machine File System (VMFS) or physical storage arrays. They increase in size “on demand” as data is written inside of the operating system. You can set a maximum size for thin disks when you create them, which is reflected by the partition size inside of the VM.

It's very important to continuously monitor your datastore usage. Thin provisioning virtual disks allow you to “over provision” your datastores, meaning more space is allocated to the VMs than is actually available. As data is written to the virtual machines, datastore usage will also increase. By default, alarms will be triggered when datastore usage reaches 75%. Storage vMotion can then be used to move virtual machines to other datastores with more available space.

vSphere & vCloud Director: Delete Unneeded Snapshots

Snapshots can be a very convenient way to preserve the state of your virtual machine at a specific point in time. While it may sound desirable to keep multiple snapshots in place, it can add up to a lot of wasted space as well as negatively affect the performance of the VM. Instead, we recommend using a true backup solution to protect your data going back later then 24 hours.

When deleting a snapshot, data from the primary and the delta virtual disks is merged back together. Depending on the age of the snapshot, that can be a very resource intensive operation and take up to several hours to complete and negatively affect performance, so you may want to perform this outside of peak hours.

White Paper

Ready to Launch Your Cloud?

Talk to an infrastructure consultant today.

Horizon View: Enable Disk Reclamation and Linked Clones

When running virtual desktops with VMware Horizon View, user data can quickly fill up allotted storage. With vSphere 5.1 and higher, the default “space efficient” disk format allows vSphere to reclaim unused disk space in linked clones. To use this feature, you must enable space reclamation for vCenter in View Administrator. Reclamation will not work on full clones or persistent disks.

Linked clones themselves are a solid way to reduce storage costs compared to persistent desktops, lowering storage use by 50-90%. View Composer treats a single “gold master” VM as the parent image for up to 1,000 VM clones. Each of these linked clones shares its virtual disks with the gold master, but appears as a completely independent desktop as far as the user is concerned.

Another option for linked clones is setting up a separate virtual disk for disposable storage, such as the paging and temporary files generated during each user session. This disposable disk is deleted when the desktop VM is powered off.

You should be well on your way to regaining some disk space. Make sure to properly backup or suspend VMs before proceeding with any of the above methods. Know of any other tips to the efficiency of virtual machine storage?