It's a good question: What ever happened to VDI? It's a question that I think some of us have forgotten to ask over the past six or seven years. I remember a time when predictions were in the billions of dollars related to converting standard desktops to virtual ones via VMware, Citrix, and others. Funny thing is it hasn't really happened. I wrote on ZDNet, in Linux Magazine, and other places as well that VDI is too expensive, doesn't perform well, users hate it, and it's a transitional technology. My last assertion that VDI is a transitional technology is really the focus of this post.
Virtual desktop infrastructure (VDI) dredges up a host (no pun intended) of images (again, no pun intended) where users connect to their virtual desktops via smart terminals at their desks. That's the picture of "traditional" VDI. But why bother setting up VDI for internal users at all? It seems silly to cripple your desk-bound users to a terminal that is neither portable nor particularly smart.
Traditional Virtual Desktop Infrastructure Architecture:
- Virtual machine hosts clustered for HA and DRS.
- Virtual desktop images. One image per user.
- SAN - Multiple tiers of storage
- High speed network - Switched Gb Ethernet or faster
- End user devices setup to access the virtual desktop image
- Management software
- Frustrated users
There are also existing solutions that leverage a single image that's replicated or copied for all users.
The common factor in all traditional VDI solutions is the attempt that administrators make to deliver a consistent desktop environment to users. The greatest hurdle to working like this is that local user's desktop performance is lackluster at best.
Take a look at why VDI doesn't provide the promised performance boosts nor the promised cost savings.
- Cost of infrastructure: virtual hosts, licensing, network upgrades, storage, end user devices.
- Cost of training users to work in the new environment.
- Management software costs.
- Administrator training.
- Administrator labor.
- Lost user productivity.
What has happened to VDI is that it's far too expensive for the return on investment. I've heard many stories of companies that attempt a VDI solution, spend perhaps five times what they would on a traditional desktop support solution only to abandon it as a failed project.
I've repeated the assertion that VDI is a transitional technology. You might wonder what I mean by that statement.
By transitional, I mean that VDI is a great idea, in theory. In theory, everything works. The lofty goals of VDI are certainly worthy of discussion and speculation, but in the real world, it just doesn't measure up to the hype surrounding it. So, what is the answer to the problem of managing desktops more efficiently, less expensively, but still maintaining control?
The solution goes something like this:
When you're "offline", your local operating system takes effect. You save your data to a local folder that's synced to the cloud when you're online. This occurs in the background. Think Dropbox. When you're online and at work, you use a secure web browser connection to use your corporate desktop. This allows a user to work where ever she is and on any device that she uses.
Of course, those devices with cellular connectivity can always connect to the corporate desktop.
But traditional VDI is too heavy and it doesn't solve performance problems. In fact, it doesn't really solve any problem. It isn't less expensive. It isn't easier for users. And it isn't more agile in any sense of the word.
The web-based desktop is the solution. I'll have more on that in a later post.
The VDI writing has been on the wall for some time now. If you don't believe me, check out these third-party analyses of the topic and the technology:
Has VDI Peaked? A Change in the Adoption Drivers Sheds New Light, and New Life
VMware still very skeptic about the maturity of VDI
Here's the single reason why enterprises aren't flocking to DaaS
So, the answer to the question at hand: What ever happened to VDI? It's becoming fossilized right before your eyes. Stay tuned.