Single-instance storage makes a lot of sense in a VDI environment. Common blocks of data can be deduped, shared and cached by VDI storage systems. In this video, VDI expert Brian Madden explains how single-instance storage works with VDI. Watch the video or read the transcript below.
More on VDI storage
What you must know about VDI storage requirements
You've all heard this thing about how humans and chimpanzees have 98% identical DNA. There's only 2% of a difference between us. We look very different, but 98% of our bits are identical. So when we think about this in the desktop world … all of our VDI users, who are all very different -- they all have their own VMDK files, or they all have their own VHD files -- we think they're all completely different. We look at them, and this is the different crazy LAN that we have to support, but how different are these users really? They're all running Windows 7, they're all running the same service packs, they're all running the same version of Office, they're all running the same version of iTunes. If we dig into it, we may find a similar ratio. We may find that we can take our entire user population of a thousand or thousands of users who all look so different, and if you dig into the files and the blocks, there's a lot of similarities. We may find that our users are 70%, 80%, 90% identical. Especially for our home drives, and desktops and "My Documents" -- [they] are redirected somewhere else and not in our images. Really, what's different per user? The few apps the user happened to install (by the way … if three out of a thousand users install the same app, those blocks are the same, too, somewhere in that system). So what we've seen now is that … we have storage systems that can take completely random, unstructured data files, and find the blocks that are identical, and consolidate those. I don't want to call it dedupe because, to me, dedupe is like a capacity, nighttime, batch process thing. I'll call it single instance storage -- block-level, single-instance storage. [This is] where we could literally take a desktop, [physical to virtual] P2V that desktop, and … on user one, [make] their 20G hard drive a 20G desktop image, that's fine. We P2V user two, and [for] half their content we have a block already, and we're not adding more to their capacity. And, as I said, the blocks that are shared can be shared and cached by our storage systems. And there's vendors here that do this today. So … VDI must be persistent. You have to have persistent images -- every user has to have his or her own image. Again, there's always exceptions to every rule, but as a matter of speaking that's where I'm going.
I want to be able to take unstructured, random data out of single-instance block-level storage [and] cache that -- I wouldn’t even call it cache, just make it fast, because lots of users need it. And by the way, there's a new project that I've started. … It's called "VDI Weekend." Because if you have a VDI environment where your storage supports single-instance block-level storage across all your users, you can now P2V your existing desktops into your VDI environment. You can do VDI in a weekend.