With 5 minutes on the clock – it’s time for another "Feature In Five"! If you joined us last time, we video-profiled SQL Cluster on vSAN; in this edition, we’ll get the inside scoop on vVols Space Reclamation. Let’s get started!
Today, our experts – Cody Hosterman and Jason Massae – take a closer look at vVols Space Reclamation and how simple it is moving from a VMFS datastore to vVols. As you can see in the video, our experts are looking at a Windows virtual machine (VM) that is filled with data, almost 400 gigabytes to be exact. Enter vVols, since one of the benefits we’ve seen is that this space, the old capacity inside a Windows or Lenox machine is stranded. Meaning, if you were to delete files, or move them somewhere else, the underlying layers wouldn’t know. Our experts are going to show you what happens when you delete these files. Now that the files have been deleted, space has been cleared and there’s no reason for the array to store this data anymore. It’s time to optimize.
In this demo, when Cody and Jason try to optimize, it doesn’t work. This is because the virtual disk is thick and tells the guest OS that it does not support UNMAP nor can it reclaim anything. So the data sits there, stranded, for a very long time. From a Pure Storage perspective, the virtual disks are always thin, always support unmap, and there is no allocation penalty on performance. That is the main benefit of vVols – you get all the cool and interesting of a thin disk, but the performance benefits of a thick disk.
Now, even though the data has been deleted, nothing was reclaimed and it still shows the full amount of space being used. So, let’s move it to vVols. And what is the process of moving something to vVols? Storage emotion – it’s simple!
With the VM being moved, it is also going to offload that process via x copy. This means that it is an array-based move, even though it’s not vVol to vVol. By restarting the VM, this process kicks in to let Windows know that it needs to relook at the virtual disks, volumes and file systems to see if it supports UNMAP. When it comes back up, Windows will realize that the virtual disk is thin now, meaning it can run UNMAP and reclaim space. With the rebooted VM, Cody and Jason can verify that the capacity is cleared out. To double-check, head over to the disk optimizer and see that it now supports UNMAP, and can run optimization because it is reported as a thinned, provisioned drive.
vVols makes it easy – from one datastore to vVols, we’ve seen customers can get 30% of their array back from VMFS to vVols because it opened up UNMAP. Learn more about vVols Space Reclamation and all of its many benefits in just five minutes by watching our video! Thanks for reading and watching, we’ll see you next time.