Last October, I wrote an article "Desktop Virtualization is NOT about cost savings or saving money." In it I argued that desktop virtualization was really about providing new and awesome capabilities for users, but that doing so would cost more than not having them. I also argued that those who thought they were saving money with VDI were actually saving money with something else, not the VDI itself. I used an example of a person who replaced $1,000 PCs with $700 worth of VDI and thin clients per user. That person thought he was saving money with VDI, but I pointed out that his new environment was much less powerful than his old environment. So while it's true that it was cheaper, it wasn't cheaper because of VDI—it was cheaper because he cut down the user experience. (8 cores per user to 8 users per core)
A lot of people agreed with that article, but some people said that I was still missing the point. They argued that I was only focusing on capital costs, and that even if VDI cost the same to build, companies could save OpEx costs with the ongoing operation. So, they argued, I was wrong and VDI was actually about saving money.
So this article is my reply to them. Today I claim that VDI is not about saving on management either.
Why? It's simple.
Most people argue that VDI is cheaper and easier to manage because multiple users can share a single disk image. If you have 100 users, surely managing one single disk image is much easier than managing 100 separate images, right?
Yes, I agree with that. I agree that managing a single disk image is cheaper and has lower OpEx than managing a lot of individual ones. But does that mean that VDI is cheaper? No so fast!
The only way that you can actually have 100 users share a single disk image is if you lock down the images (since individual users can't install software) and implement some kind of user virtualization, app virtualization, and/or layering. So when you do all that, then yeah, 100 users can share a single disk image and that will be cheaper.
But here's the thing. If you do that, your cost savings comes from the fact that you take a Wild West "before" environment where any user can install anything and replace it with a formal, well-designed, locked-down environment where users don't have the same freedoms as they did before. In other words, your savings isn't coming from VDI at all—it's coming from the fact that you're taking away your users' freedoms.
Cheaper? Yes. Because of VDI? No.
If you just want to save money, why don't you just take away your users' rights on their existing traditional desktops? You don't VDI to do that at all! App virtualization, layering, and user virtualization all work well on traditional desktops and can make traditional desktops easier to manage.
More VDI hating?
I'll reiterate what I've had to defend about 100 times in the past few months. I don't hate VDI. What I hate is when VDI fails because it's used in situations where it doesn't make sense. I hate when people try to take their existing Wild West environment and formalize it and fail, and then blame VDI.
I'll be very clear. VDI is awesome. You can use it to allow users to work anywhere on any device. You can use it to provide the ultimate reliable desktop. But these awesome capabilities come at a price. You can't expect to have all these awesome new features and to pay less money than you pay now. That's just not the way the world works.
VDI means central VMs. Surely that's easier to manage?
Many people know that I love 1-to-1 VDI (where each user's disk image is individual) because that's the most like "normal" desktop computing. So if we're talking about 100 unique users out in the field somewhere versus 100 unique users in VDI, wouldn't the VDI users be easier to manage? Even if you have to use SCCM or Altiris or Windows patching, wouldn't you much rather do that to 100 VMs in your data center rather than 100 laptops out in the world somewhere?
But again, making these desktops easier to manage is not the primary reason for going to VDI. VDI is expensive. It requires a major change in the way that users work, and a lot of things that users could do before just won't work. So in this case, I'd argue that the 100 VMs being local is a nice bonus of VDI, but it's certainly not a reason to go to VDI. If you just wanted those 100 random machines to be easier to manage, I'd probably go with something like VIrtual Computer NxTop, Wanova, MokaFive, etc. Way cheaper than VDI. (Because again, VDI is not about saving money—it's about enabling awesome new capabilities.)
I love VDI. It's just not about saving money *or* about making desktops easier to manage. Yeah, there are a lot of things you can do to make desktops easier to manage, but you don't need VDI to get them.