As most people know by now, AppSense received a $70m investment from Goldman Sachs. I was surprised as anyone as to why Goldman would make such a substantial investment in the user virtualization space.
We've had Windows mandatory profiles, roaming profiles, and folder redirection for years. But no matter what Microsoft claims, all of us practitioners know these are very limiting. So you'd think everyone would be falling all over themselves to implement a third-party solution such as Appsense, triCerat, Scense, etc. While many people do use these add-ons, I still don't consider them as a requirement in the stack. Why? Read on...
I always assume most people cite "cost" as the reason, but I'm skeptical that's the real reason if you look into it. I think the real reason is that most people are still in highly homogeneous unmanaged environments. Most people still allow dumb things (like admin rights on PCs) and then hope that user-installed apps will be a miracle cure (which it will never be). I'll bet those same people just use local profiles and clear them when they become corrupt. I bet their users don't roam and they work at classic organizations where people are tethered to a single desk. I bet the "rebuild the PC/profile" mindset is the SLA at these organizations... and all that craziness is just the profile part! If you look at other areas, I'll bet things like login scripts, group policy management, application management, etc. are a mess and not very well managed. (And this is made worse over time as more people enter and leave the organization.)
So for all these poorly-managed organizations, the next unfortunate thing they assume is, "I need to bring my management costs down, so I'm going to virtuale all my apps because that's what Microsoft says I need to do and it's what so many people blog about." :-) So they start down this path and realize application virtualization is not perfect (and they freak out when they learn they need MDOP just to get App-V). Others go for ThinApp until they realize there's no management and it's not a silver bullet either. So then they think surely this "user installed apps" thing they've been reading about is the real silver bullet and they start looking at some of the startups in this space (only to find out they don't necessarily scale, perform, avoid application compatibility issues, or work across platforms).
Then, these same people think, "Ok, desktop virtualization must be the answer since the vendors are telling me it's cheaper." Or perhaps they read some other website which has rivetingly informative articles that tell you exactly what the vendors want you to hear. After reading all that crap the desktop teams still don't understand that VMware lied about SBC and still don't have a clue about how VDI is different, and Citrix keeps saying Receiver will support tablets in outer space if you want so that's pretty cool... Then they read Brian Madden's blog and think, "Wow there are smart people here," and they see a debate on SBC vs. VDI and think they're the same thing because some commentators said so and they just go back to their day jobs thinking, "I should keep trusting these people to inform me of basic tenants that I MUST understand. "
Once they finally understand what "hosted desktop virtualization" means, they realize a few things. First, to get the cost savings with VDI, they need to implement a single image model (once they swallow the licensing cost pill). Of course when they try it they realize it doesn't work well today outside of a few silos. They then wake up and say "Ah-ha! Now I understand why SBC is not VDI and can achieve the savings over a PC." Then they sign up with Citrix or Quest (if they don't understand or care that RDP sucks and Microsoft is using RemoteFX to lock you into Hyper-V despite your investing in ESX) and go down that path only to realize that even with SBC there are limitations. (For example, personalization is limited, and they have to manage RDS, desktops, and laptops.) Then they realize it makes sense to probably support multiple models for different users due to the economics of the various models. In their minds this is even more complexity that they have no time or budget to deal with. So what do they do? They just stay exactly as they are, implement some application virtualization, and migrate status quo to Windows 7. Microsoft has accomplished their mission of keeping Windows on the endpoint for another OS refresh while they try to figure out WTF is cloud going to mean beyond a huge advertising campaign that says it's more than virtualization.
Despite all of this, many people will become successful with desktop virtualization. They'll implement SBC in a highly managed form with next to zero personalization and achieve costs savings for many use cases. Others will implement 1-to-1 VDI using exactly the same management processes and tools as today, touting benefits of centralization, agility and flexibility with some cost increases justified by new capabilities such as Windows apps on tablets. Most of these people will end up with a mix of SBC and VDI -- both managed differently -- and two sets of application validation test platforms. Add laptops / client hypervsiors to the mix and there's yet another environment to manage. All this will keep desktop virtualization to only a few percent of all enterprise desktops which is in-line with the various expert reports I've seen.
Can you believe it? Only a few percentage points of total desktops! Why are we spending so much time talking about this space if it's not relevant to the majority of the desktop community? This is why I was scratching my head at the Goldman Sachs investment. Then it dawned on me after some digging... Appsense is arguing that user virtualization is about all users being managed. That applies to physical desktops as well. They offer more than profiles and can go cross platform. They even seem to be trying to help the App-V team. I tweeted Ruben Spruijt to ask Microsoft about their latest vie. So if you can get your users under management then you can abstract all that complexity away and leverage it across platforms when you see fit. To get the cost benefits of VDI or SBC or any other single image desktop virtualization solution, you must have a managed desktop. So why not start with the user as opposed to the application? Better still, do both at the same time. Just getting to a better managed desktop state will reduce cost. So my guess is Goldman looked at this and said the VDI market by itself is $2B over the next few years. If we invest in something that can appeal far more broadly to desktops then what the heck is that worth? $70M is probably just a drop in the bucket. Imagine if Microsoft made something like this part of MDOP and took away cost concerns despite my assertions above and made any Windows desktops or application cheaper to manage. Wouldn't that be a great goal for them and win for customer? What if VMware picked up Appsense and made it part of VSphere? How powerful would it make them as opposed to messing around with layering startups? What if Citrix added this to their portfolio to make it easier for people to use all their desktop models? What about Cisco UCS having this as part of their base solution to help drive down costs, or even somebody like a BMC adding this to their portfolio? This stuff works now, and layering technologies may never work. This to me must be the logic behind the investment.
I may or may not be right with my assumptions, however whether you love or hate Appsense or any other personalization vendor, to adopt a greater variety of desktop models with a shift to a better managed desktop is a must to get the TCO down for all desktop models. Part of that has to be user management. I'd love to know what the community thinks? (And vendors are free to comment as well.)