Is deduplication an oxymoron? It is with VDI!

I’ve been thinking about deduplication recently. The very name strikes me as odd. De-duplication. Do something so that we can undo it. Create duplication, and then undo the duplication. Does that make deduplication an oxymoron? Here's what my dictionary says:

ox•y•mo•ron (noun) a figure of speech in which apparently contradictory terms appear in conjunction to an incongruous, seemingly self-contradictory effect (as in “cruel kindness” or “to make haste slowly").

OK – maybe not an exact match. But nonetheless, why would we want to spend the money, compute and I/O cycles to create duplicates in the first place just to spend even more to undo them? Case in point: Deduplication of virtual desktops to reduce the cost of VDI.

It logically makes sense – we can reduce the total amount of storage required, and therefore cost, by finding all those duplicates that exist in desktops. Every desktop has a copy of Windows, and likely has Office, and countless other applications. Commonly the “gold image” of Windows and applications can be 80% or more of the total desktop storage space, resulting in massive reduction of storage if deduplicated.

Just think – a 20GB gold image and 2GB of user data, multiplied over a hundred desktops, would take 2.2TB of storage. But deduplicate it and you now have 220GB of storage – 10 times less. So what’s wrong with that?

What’s wrong is why allow the duplication in the first place? Why not have just one copy of Windows that is shared by all users? And one copy of each application? And isolate everything the user does, including data, settings, add-ins and personal applications into their own layers?

It's much easier to manage, since you can now patch and upgrade the one image and share it across all desktops while retaining all personalization. It also eliminates duplication, and therefore the need for deduplication – as well as all the cost and overhead.

I think it makes sense. Of course, I'm biased, since this is exactly what Unidesk Composite Virtualization™ technology does for VDI. It naturally eliminates duplicates to begin with by isolating Windows, from the applications, from users – and lets common images be shared across all desktops. It eliminates the need to use deduplication in order to achieve massive storage savings.

If you want to learn more, take a look at our videos on Composite Virtualization, our dynamic image composition and decomposition technology, and CacheCloud, our delivery technology for storing, replicating, and versioning those images.

Deduplication. Still feels like an oxymoron to me!

-Chris Midgley
Unidesk Founder and CTO