In our "Trends for 2010" episode of Brian Madden TV, a few of the trends we focused on included a "renewed client-based VM focus" (thanks to the release of client hypervisors from VMware and Citrix this year) and a "smarter replication" that will replicate just the changed bits of files that actually matter instead of trying to "check out" an entire VM image. (I wrote a fair amount about this concept in the "Think of the awesome ways that RTO could help VMware’s desktop strategy" section of my article about VMware & RTO last week.)
In the past few weeks, I've become really excited about this idea. So excited, in fact, that I'm going to go on record and say that 90% of desktops in the future will run locally on client devices. (Be that via a Type 1 client hypervisor or a Type 2 client-based VMM.)
Remember, VDI is just a form of server-based computing. As such, it has the same advantages that server-based computing has always had, namely centralized management, access to Windows apps from non-Windows devices, good performance of multi-tiered apps over WAN connections, and "eyes only" security.
Many people implemented Terminal Server-based solutions to get these server-based computing advantages, and that was fine. Now that VDI is an option for people, some traditional desktop customers are implementing VDI to get these same server-based computing benefits. (And some Terminal Server customers are converting over to VDI.)
But the vast majority of the user base out there doesn't need VDI. Sure, they need some of the benefits of VDI, like ease of management, secure computing, and access from anywhere. But does that mean they need VDI? Maybe if they want to do this today, VDI is their only option. (Well, the only option apart from just continuing to do things the "old" way.) But when client-based VMs become real, that might be the better option for these users.
Client-based VMs will leverage the power of local computing, enable offline, allow great local performance, require smaller datacenter footprints, provide for centralized management, provide for easy backup and rollback, etc.
So when client-based VMs become real. (Sure, companies like MokaFive are real enough today, but the underlying OS is a huge problem for them.) When Type 1 client-based VMs become real, we get the central management of server-based (Terminal Server or VDI) with the flexibility of local. Over the next few years, many traditional desktops and laptops will move to client-based VMs.
Does that mean that Terminal Server and VDI are dead? Of course not. Terminal Server provides a great low-cost desktop for simpler workers. Terminal Server also provides a really great way to deliver single applications that specifically require server-based computing. (Maybe you need the "eyes only" security. Maybe it's a three tier app and you need the Windows fat client to run right next to the database.) Terminal Server is not going away. It's just that in the future, it will be used to tactically deliver a few apps here and there.
Same for VDI. Last month I wrote that everyone who really needs VDI already has it. And I honestly believe that. It's crazy to think that huge swaths of people will use VDI who aren't already. What's real is that huge swaths of people want the manageability of security of VDI, and soon they'll be able to get that with client-based VMs.
Actually, these client-based VMs might enable more VDI in the future, because if they're done right, they'll be continuously in sync with some central host, so it'd be just as easy for a user to connect to their desktop via VDI as it would to connect and run it on a laptop. But that's sort of using VDI as a "free bonus" connection method to what will really be enabled by the client-based VMs.