A few weeks ago, I wrote an article about the fact that Symantec had suddenly stopped selling their Workspace Profiles product which they OEM'ed from RTO Software. I wondered why this was, and suggested that perhaps it was because Symantec competitor VMware was planning to buy RTO Software.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Yesterday both VMware and RTO Software announced that was true (via Warren Ponder, Scott Davis, and Chris Westphal—Yay bloggers!). Actually, I should say that's "almost" true. Before yesterday, RTO Software had four products:
- Virtual Profiles
- PinPoint (performance monitoring)
- Discover (for discovering what's out there on your network)
- TScale (memory, CPU, and registry optimization for Terminal Servers)
But VMware just bought the first three products, leaving TScale with RTO. (And if you go to rtosoft.com, you'll see their whole website is just about TScale now.)
Why doesn't VMware want TScale?
Before we dig into the analysis of why VMware bought these products and what they'll do with them, let's address the easy question first: Why didn't VMware just acquire the whole company?
In my mind there are two simple reasons for this: Remember that back in 2004, RTO Software licensed their TScale product to Citrix for inclusion in Presentation Server. (You can read our coverage of that 2004 agreement here, because hey, we were blogging way back then!) In my August 2005 audio interview with RTO Software founder Kevin Goodman, we learned that RTO had an operational agreement with Citrix where RTO would support the TScale product for a number of years. And since Memory Optimization Management is still a feature in XenApp 6, we can assume that RTO is still supporting it. And I can't imagine that VMware wants to have anything to do with supporting Citrix here! Couple that with the fact that VMware doesn't have a real solution for Terminal Servers (which is the whole point of TScale), and you can imagine a VMware exec saying, "Ummmm.. Yeeeeah... TScale? No thanks!"
So instead of buying the whole company, VMware buys the whole company minus TScale, which is why RTO Software is once again "The TScale Company!" RTO's Kevin Goodman will join VMware as a director of product management, which is great for him and VMware. Anyone who's been to BriForum knows that Kevin is one of the smartest and the nicest guys around.
What happens to RTO Software now?
RTO Software began life as a single product company (around TScale). Now that RTO Software is just TScale once again, the remaining employees from the TScale group were actually able to use the proceeds from the sale to buy the company. So now RTO Software truly is "The TScale Company." The remaining folks at RTO are really excited about this, as they feel that they're able to get back to their roots and they're happy to be able to do what they want.
In addition to TScale for Terminal Server, RTO has been talking about TScale for Hyper-V, which we're definitely going to hear more about in the coming months. (I guess that could be Reason #3 why VMware didn't want TScale.) Kevin's going to stay on as interim CEO for a few months until they can find someone to run RTO full time. (This will be in addition to his VMware role, so he's going to be a busy guy for awhile!)
What does this RTO deal mean for VMware?
When I first wrote about a potential VMware / RTO deal earlier this month, I was excited about how the real value of RTO Virtual Profiles was deeper than the Windows profile, writing:
When RTO Software was our Random Vendor of the Week on Brian Madden TV last May, developer Eric Tatum showed us some new functionality where the Virtual Profiles product could be used to manage and sync files outside of a user’s profile folder. That’s what ultimately led me to make the prediction from last week’s Brian Madden TV about 2010 Trends where I said that VMware has to get beyond this whole concept of copying and replicating entire VM disk images.
That’s the problem with VMware View’s experimental offline mode right now. When the user clicks a button to take his or her VM offline, View then copies the entire disk image (or disk image delta file) down to the user's workstation. But since that’s happening at the virtual hardware level--below Windows and below the file system--90%+ of the blocks that make up that disk delta file are completely worthless and do nothing but increase the sync time.
To further understand this, think about what happens if you just boot up Windows, make one little change to a text file, and then shut down Windows again. From a user perspective, you only need to change a few bytes of data in your text file. But how many blocks on the virtual disk actually change? Probably hundreds of megabytes by the time you look at time stamps and temp folders and the page file and all the other stuff that Windows writes to the disk as it’s used.
So VMware’s current Offline VDI concept is this “dumb” mode where it would have to replicate lots of megabytes to the client in order to transfer down the 1k file change. But if VMware owned RTO Software, and if Eric Tatum & Company could rebuild VMware’s offline syncing to happen “in band” (like “inside of Windows”) at the file (or part of file) level, then VMware would end up with a pretty bad-ass solution!
If done right, a user wouldn’t ever have to “check in” or “check out” a VM. The user’s data would just keep syncing in real-time (like with Dropbox or Live Mesh), so prepping the VM for offline use would be a simple as logging out and logging in to another location. (Or maybe VMware could combine live migration & LufLogix to just make this “flow” from device to device?)
I can't put enough emphasis on how important this is, especially since I think 90% of future "VDI" users will run VMs locally on their clients.
In fact I was so excited about this concept (and so sure VMware would buy RTO) that I really pushed this concept when I interviewed VMware desktop CTO Scott Davis a few weeks ago. Unfortunately he didn't seem too interested in this approach, instead insisting multiple times that disk data can be intelligently replicated at the block-level from outside of Windows. (I can't say that I agree, but whatever... There's a reason this guy is CTO and I'm just a blogger.)
Remember, VMware's getting more than just profile virtualization
Of course this VMware / RTO deal is about more than just RTO Virtual Profiles. VMware is also getting RTO's PinPoint and Discover products. If you read the acquisition FAQ (PDF link), VMware plans to discontinue the RTO stuff as standalone products and to instead integrate them into View and ThinApp.
PinPoint is an application performance monitoring tool used to inform admins of potential problems with virtual desktops (slow logons, apps that are taking forever, etc.). One of the cool things about PinPoint is that it's focused on the process of why an app is slow as opposed to the microscopic milliseconds of data. What I mean is that most performance monitoring tools look at things like clock cycles to figure out how long different steps take to complete. But in a VM, the cycles in a guest do not necessarily match up to the physical real-world processor cycles. So since PinPoint is designed to run in a guest, it actually ignores process cycle times and instead looks at the time of day clock. This means that while PinPoint might only have accuracy to the nearest second, it can safely understand physical and virtual machine timings.
What's cool about this is that PinPoint bases its analysis on these real world timings. So for example if PinPoint is showing that a desktop login is taking two minutes, it can show that the AD connection took 10% of that time, and that the logon script processing took another 38%, etc. This means that PinPoint is not the tool to track SLAs for customers or users, but instead that it's meant to help admins figure out where the problem is.
PinPoint is also one of the few products that can look at the performance metrics of the client device (if it's Win32) when the user is connecting via RDP, ICA, or PCoIP. It can actually show an admin that a user is doing crazy stuff locally on their client (as opposed to inside their VM) which might negatively affect performance.
I would imagine that PinPoint will fit nicely into View to be VMware's in-box desktop performance monitoring solution. Maybe they can also use it to monitor the performance of ThinApp virtualized apps? (Which will be getting faster in the next release, by the way!)
The final RTO product that now belongs to VMware is RTO Discover. Discover is sort of a "down and dirty" asset discovery tool. You install it on a single admin workstation and let it run, and it goes out and finds and inventories workstations, applications, servers, etc. It's not meant to be a full on asset management system, and it doesn't have the intelligence of Liquidware Labs or Lakeside Software's products to make recommendations about how your future VDI environment should look. But if you want a simple tool that can be done in 3 hours instead of 30 days, Discover is what you want.
I would imagine that VMware would maybe provide Discover for free (maybe just to partners?) to help people understand where their desktop starting point is before attempting virtualization? Again, VMware doesn't have anything like this now, and Discover is very easy to use, so it's a good fit.
What about others? Symantec? Citrix? AppSense?
So pretty much everyone thinks that this RTO deal was a good move for VMware. (Well, everyone except for VMware's competitors. How are they handling it?)
Ever since I reported that Symantec has stopped selling their Workspace Profiles product, I have not actually been able to get anyone from Symantec to share their thoughts on this. The same is true today. I still haven't heard a peep from Symantec about this despite asking a few different people.
Symantec isn't the only casualty here though. AppSense's Martin Ingram wrote a passive-aggressive "Congrats VMware and RTO" post, where he basically says that RTO's (and now VMware's) base-level personalization is fine as a starting point, but that it's not a real enterprise-class user environment solution. Martin was also critical of the "fundamental problem in a user virtualization solution that only works across a single vendor’s products." Presumably he's inferring that VMware's profile virtualization will only work in View environments, so what are customers supposed to do for their offline desktops, Terminal Servers, etc.?
It's a fair point. VMware would probably respond by saying that they soon expect every desktop to be in a VM (whether local or remote), so that issue will soon be moot. (I'm sort of half-writing that as a joke, but I know there are high-level people in VMware's desktop group who truly believe this, so we'll see.)
Of course you've got to consider the source of the "single vendor, single environment" critique. Even though AppSense does a lot more than just the user profile on VDI, every time a little "profile only" product is snapped up by a big company (Citrix acquiring Sepago Profiles, VMware acquiring RTO Virtual Profiles, etc.), companies like AppSense and RES Software have a harder time selling their stuff. (Although all's not lost. I hear Symantec might be looking for a new user environment solution.)
(And God-willing, so will Microsoft. :)