Correctly calculating the storage space required for your virtual snapshots can be tricky to determine and can have devastating effects if done incorrectly. If your snapshots exceed your storage capacity, your virtual environment will go down and you will have a very bad day.
This is particularly important for Veeam users since it automates snapshotting.
Just as a background, snapshotting requires extra space allocated because for the duration of the snapshot being present, the entirety of writes that come to the VM are stored in it. The original disk doesn’t change anymore, so rewrites as well as net-new writes are both stored in the snapshot disk.
In order to avoid this type of pain, here are a few things you’ll want to consider when allocating your storage.
Understand your placement of VMs on your datastores
Since VMware can now support larger datastores and VMDKs, it’s become increasingly important to track the individual snapshot space of what a VM would require.
It’s easier to track your snapshot size potential if you have it set to be one VM per datastore, but that can cause issues with management the larger you get.
Now that datastores can be up to 64 terabytes in size, you’re likely to have a lot more VMs on one. Therefore, all of those VMs snapshotting at the same time can take up a lot of storage space and impose overhead on the hypervisor, which makes it more difficult to predict.
Account for VM workloads
Not all VMs are created equal. Some might have high rates of data change, like database or email servers, that typically have much larger snapshots than, say, a web server in which the information is essentially static. Because some VMs require more space than others for snapshots, you can’t do a simple one-size-fits-all rule of thumb when calculating your required snapshot storage allocation. A good place to start, however, is to leave at least 20 percent of the whole datastore’s size allocated for its snapshots.
We also recommend spreading out your high change rate servers amongst your datastores in order to keep a more steady average.
Documentation is your frenemy
Nobody likes to do documentation. It’s tedious and can require regular updating. However, when done correctly, will not only save your hide during a disaster but it will help you with asset management and storage tracking.
This includes documentation of workloads and change rates so that you can make educated decisions on how much storage to allocate for each snapshot. One tool I like to use that can help you run an inventory on your virtual environment is RVTools, which is free.
Consider using monitoring and/or assessment tools
The most important thing, even without any paid tools, is that you can set up monitoring and alarm notifications within vCenter that will notify you when a snapshot is over a certain size or when a datastore is below a certain percentage of free space (which you configure) and that will help you stay ahead of any problems.
There are all kinds of different monitoring and third-party applications that will help you monitor, track, size and automate your virtual environment, including snapshots and datastore management. These programs require you to configure your thresholds and alerts, but are built to help you proactively manage your environment.
You can also have assessments performed on your environment, which can help you inventory and baseline. If you’re in a new job or think you need to check up after a certain amount of time, these assessments can shed a lot of light about your environment, including VM placement and workloads.