VMware licenses RTO’s Profile Management technology for use in future VDI products

The only actual new announcement that VMware made about desktops at VMworld last week was that they were licensing RTO Software’s Virtual Profiles product for inclusion in some future version of VMware View.

The only actual new announcement that VMware made about desktops at VMworld last week was that they were licensing RTO Software’s Virtual Profiles product for inclusion in some future version of VMware View. (press release)

This is great news, as the whole “user data disk” plus roaming profiles technique currently employed by View doesn’t really cut it in the real world.

If you’re not familiar with RTO’s Virtual Profiles product, check out Cassondra McAllister’s demo of it from the DEMO Lab at BriForum 2009 this past July. (The part on Virtual Profiles starts at 1:50 in)

This is an OEM deal, meaning that RTO will maintain the codebase and will continue to sell the Virtual Profiles product on their own, but that at some point you’ll be able to buy VMware View products that include RTO’s technology. VMware has not yet announced any details as to when this might be available.

RTO’s founder and Virtual Profiles creator is Kevin Goodman. Kevin’s been a presenter at the past six BriForum conferences. He co-created the “Citrix Logon Process Chart” with me and he was our very first guest on Brian Madden Live! (Our podcast which has been replaced by Brian Madden TV.)

On the business end, Kevin created the TScale DLL optimization product which he licensed to Citrix (which became the “Memory Optimization Management” feature of XenApp a few years ago). This past March we learned that Kevin licensed his Virtual Profiles technology to Symantec, which is the same core technology he just licensed to VMware. So congrats to Kevin!

Kevin was at VMworld last week and I pulled him aside for a quick interview which we put into last week’s episode of Brian Madden TV. (Interview starts at 2:54)

Why RTO?

With so many profile virtualization / user environment management products on the market, why would VMware choose RTO Software?

Of course we don’t know what the back room negotiations were like, or whether VMware shopped around between several vendors, but one thing that’s interesting about RTO’s Virtual Profiles is that it doesn’t leverage a back-end database or management system. It’s just an agent and some data stored in the profile folder. This is very similar to how ThinApp works and something that makes both ThinApp and RTO different from their competitors. So while it’s not that this is a better technique per se, it’s interesting that both products work in this way and it makes sense that VMware’s would include products that are both similar in this way.

Is this “just” profile management?

Two years ago I wrote a blog post wondering whether it was possible to “just” manage profiles or whether we needed something the manage the entire user environment. That post was based on a conversation with AppSense’s Pete Rawlinson, and Pete actually posted his thoughts to AppSense’s blog last week about the RTO announcement.

Pete argues that while this OEM agreement is a good thing, RTO’s technology only focuses on profile management and doesn’t capture everything that’s needed for managing the “complete” user personality. (So again, this is the same conversation we’ve been having for years, and boils down to whether you want to capture changes that were only written to the user’s profile area or to the entire system.)

Pete makes a good point. The current version of RTO Virtual Profiles does only capture profile changes. That said, RTO is working on a new version of their product that can capture and restore changes written anywhere on the system. This is something that we first learned about back in BMTV Episode #16 when RTO Software was our random vendor of the week. Chief developer Erik Tatum demoed how they could, for example, capture changes written to c:\mystuff and redirect those into the profile folder. (Video here, with this portion beginning at 24:50.)

So today that’s just a future technology from RTO (although they were demoing it at their booth at VMworld last week). But according to Kevin (from BMTV #17), it sounds like VMware will also have access to that capability as it comes online. And this is critical to VMware.

RTO’s product can evolve to help fix offline VDI

During the Day 2 “technical” keynote, VMware CTO Stephen Herrod mentioned that offline VDI was broken. He was talking in the context of syncing disk images, because VMware’s current offline solution leverages the VMDK delta differential file technology. This is a problem because it works at the disk block level, so you end up with a scenario where you have to synchronize huge amounts of data with your client to take it offline, but most of the data ends up being stuff that you didn’t need anyway.

Just imagine how much garbage Windows writes to or changes on the disk. Sure you can make policies to zero out the page file and stuff, but if you’re looking at Windows from the outside, you’re going to end up moving a lot of worthless disk blocks around.

A better way would be to run “in band” so you knew what was valuable and what wasn’t so you only moved around the good stuff. This is where the new capabilities of RTO Virtual Profiles can come in. So now instead of trying to use policies to redirect user changes to auxiliary VMDK files, you could write everything to the c: drive and let RTO capture the changes and sync them in real time up to a file share. Then you’d only need to sync up the actual files to take the thing offline.

This new capability will also help with the complexities that arise from multiple users sharing a single master disk image. (What VMware calls View Composer.) Sure, you can use ThinApp to add apps on demand, and you can use the traditional RTO Virtual Profiles to capture and save profile data. But what about those changes that would be written to the c: drive and then lost when the VM was refreshed? RTO’s new stuff would allow those too to be captured.

So while we should be excited about VMware OEMing RTO Virtual Profiles for “just” profile management today, the real value and brilliance of this deal will be apparent in a few years when they start incorporating this capability into other aspects of their View product.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

I don't get why anybody would really care. VDI regardless of vendor will take some time to migrate to regardless of your penetration rate. In this mixed environment are you REALLY going to introduce yet another profile (YAP) into your environment. For the TS use case you are forced to do so, and there it's a pain also to manage YAP. Until we have leapfrog with layers to the next thing, I find it very hard to justify the effort to even bother.

Also Brian, I think there needs to be a debate on persistent image VDI vs. pooled. Who's doing which one. My view personally, pooled is not mature and is too hard right now with the technology, it doesn't scale, implementations suck, install is complicated (XD really sucks at this) and there are just loads of risks with pools. I am going all persistent day one, hence why I don't give a hoot about the middle tier and simply want HDX Connect to meet all my use cases. I will invest in pools over time, with layers and then bother going through the effort of engineering my systems management tools. Windows 7 will help drive that, but for now I don't plan to mess around with profile management solutions in a mixed environment since I have no idea what's next that will likely cause me to do it all again anyway, plus pools SUCK.

The one thing I am 100% with VMWare on is the lack of DB. I find Appsense too complicated to set up and mess around with, same with RES, although I have spent limited time with them. Here Sepago with Citrix is ok, but again missing many features to be real for a VDI environment, which like I said does not matter.


If you are going to use pooled virtual desktops, physical desktops, terminal services, or if users will have any more than one of each...separating the user state from the host is critical.  

This can be done with roaming profiles, but those are inflexible, unreliable, slow and introduce way too many support calls.

We've been addressing all of these issues on TS/RDS since 2003 and ported our user profile management to the desktop (physical or virtual) in vWorkspace 6.0 (January 2009).

As for your pooled vs persistently assigned desktops, I see if done both ways with my customers.  Most users get by just fine with a pooled/termporarily assigned desktop, but some people (developers) are better served with a persistent desktop, or a few of them, i.e. dev, test, build/compile...


simple question : how is this RTO profile management compare to the regular "up to date" W7/W2K8R2 profile management and other solution ? I heard that Citrix have something called also "profiles management" as a feature of both XenApp and Desktop...


I don't buy for a second that any large customer is going to use pooled desktops. They can't get their freaking inventory right, let alone the rest of the stuff to enable pooled. Pooled will take years to implement at scale, and assumes you know what users need, which is a massive inventory problem. Only very simple use cases or organization can leverage pools today. This is where so much fud is caused by the vendors. Ask the vendors how many of their own users are using pooled virtual desktops....as the only way to work. It's fud, so beware of pooled desktops. Persistent is the only way at scale today that I see as real.


Sooner or later, it's very likely that we'll cease to distinguish between persistent and non-persistent (pooled) desktops. Of course, this is contingent on whether emerging technologies, focusing on decoupling the OS layer from the user personalization and apps layers, will succeed at accomplishing this feat. Once such decoupling is achieved, the desktop can then be recomposed on demand by combining L1 (the stateless OS), L2 (the apps layer), and L3 (the user layer). In this manner, (L1 + L2 + L3) and (L1 + L2)  would constitute persistent and non-persistent desktops, respectiviely.

Of all the vendors in this space, the one that holds the most promise (in my opinion) is Atlantis Computing. But it seems that UniDesk and others are not too far behind. However, Atlantis Computing's approach is pretty unique among others, not to mention the performance benefits realized through I/O virtualization.

If a virtual desktop management product (aka. Connection Broker) were to integrate with a solution like Atlantis Computing's ILIO sometime in the future, the three layers, L1, L2, and L3 would have to be exposed in the Broker's management console as checkboxes, specifically in the properties of the desktop group (the container of desktops), as well as the individual member desktops (to optionally override the group-level property) . The desktop administrator would then be at freedom to combine L1 with (optionally) L2 and L3, as well as discard any of the layers on subsequent logons. Additional layers, call them L4, L5, etc, representing OS patches, new apps, or existing app updates, can also be exposed and selected to partake of the dynamic desktop recomposition process.  

If and when such products eventually mature to this level, L2 will invariably become the new "profile", therefore rendering traditional roaming profiles obsolete.

Without this sort of groundbreaking paradigm shift, desktop management in the virtualization age will largely remain "business as usual", and the ROIs will certainly be very hard to find.  


Correction - L3, not L2, will invariably become the new "profile", therefore rendering traditional roaming profiles obsolete.


@Peter (assuming you are the Quest guy), you can do much of that today with Moka 5 layers on the client side. On the hosted side, it will take years. Tell me which organization understand their own environment to drive layers with pools to a point where they will really trust it. I agree it can begin to emerge on the persistent model and over time the two could converge, but that's a long long time away. People have investments in existing systems management tools today, and those will not just go away. There is just too much process to unwind. Just the simple act of moving a desktop into a data center and connecting to it is a huge leap of faith. Today the brokers themselves suck, they will fail and cause outages impacting too many desktops. Folks like Citrix keep selling the pooled crap model with all the stuff you don't need, not understanding it can't be implemented at any scale, and what they have is not needed for persistent desktops. They certainly don't have product managers who get the space, all executive fluff and rah rah talk. I've spoken to them and so much crap they talk and no real world experience. Just look at the setup joke that is XD if you buy into the whole stack, which I avoid. The management team should be ashamed of themselves for allowing such a tragedy to get to market. How long did these morons take to figure out the XA plugin and VDA MUST live together. How many consoles should one manage? What the heck are the XD team really doing, HYPE if you ask me, they don't get the real world. A simple connect is all you need for a tier of users to get agility. If RDP was good over a WAN which it's not, I'd just use it. I don't get why MS just doesn not license HDX/ICA and extend it with Calista etc. Stop wasting time and own the display protocol market. I give VMware a lot of flack on this blog but Citrix are also just as stupid in the XD team. If you find a firm that will do pools at scale, tell me I want to work for them :-) That said I agree that layers is where the vendors need to invest, not this half assed let's make profiles a bit better approach. Atlantis gets a lot of hype on this site, and I have yet to see hear about real world implementation at scale with those guys. Unidesk looks interesting, but again so do the 50 other layering solutions, Ringcube is cool too. This whole thing needs to sort itself out. My fear is that many will do nothing there years and just toy with the idea because nobody else will move. VMware are treating this is a tier two space, and I expect Citrix to move real slow. Perhaps Quest will be first.......but do they have the investment ability to really make that happen? Even then they have all hassle of not owning a own protocol or hypervisor, which makes people like me very nervous to bother. I've said it before VDI (excluding XA/SBC) is not a hard $$$ cost play today, you WILL FAIL at that. It's about investing in agility and for the future making a bet that we will get there with the shift to layered cake and combining this with client side virtualization. It's a question of when people will get on that train and when and IF the technology works. For most people, VDI I expect to make little sense for years, and they will simply stay stuck in their old ways, don't get it, don't have the balls to make a strong argument, even with Windows 7 smacking them in the head. There are plenty of stupid admins out there who do not want to change or do anything. Same things happens as you go up the chain until you get to the money guys who will get it if explained right. Its also safer to do nothing and let somebody tell you to do it. The Sheep mind set is everywhere. There are very few VDI champions who get how to implement. So Persistent at scale in a simple way with existing systems management tools moving to a brighter future as the technology becomes better is the only way I can see how to do this in the real world at scale.


Actually, the former Quest guy :)  The focus of my current venture is not on the desktop industry any longer, but I do like to chime in every once in a while.

Yes, I've been tracking Moka 5 for over 2 years, but until the hosted side of VDI sorts out all of the current hurdles, most companies won't seriously start looking at bringing the whole thing full circle. Having said that, those who won't find solace in host-side VDI may come to realize that client-side VDI is the answer. Therefore, solutions from the likes of Moka 5, Neocleus, Virtual Computer, and others, will eventually start gaining traction.

Having said that, client-side VDI doesn't necessarily require a hypervisor. And layering, in general, can be accomplished in several different ways, one of which is Atlantis Computing. In my recent webcast with Brian (www.brianmadden.com/.../live-webcast-today-with-brian-madden-and-peter-ghostine-noon-edt-16-00-gmt.aspx), I pointed out that environment virtualization solutions from the likes of RingCube, InstallFree and Ceedo could very well complement a VDI implementation and extend it all the way to the client. These solutions transcend app virtualization as we've come to know it over the years, extending the concept of "sandboxing" to include the user environment itself, or what we affectionately refer to as the "user profile". Therefore, if VMware had any vision at all, they would recognize that they've already got the technology they need to build a "virtual profile" solution. In other words, all their ThinApp division has to do is to extend ThinApp to virtualize the user environment, not just the apps. And for that matter, the same goes for Citrix and Microsoft with respect to their app virtualization solutions.

Regarding to brokers, obviosuly not all of them are created equal. I encourage you to give Quest vWorkspace an objective look, recognizing that owning the hypervisor layer is not a curse, but rather a blessing, since you'd be at freedom to choose any hypervisor you want, including VMware ESX and MS Hyper-V, or OS partitioning as is the case with Parallels Virtuozzo Containers.

Regarding protocols, I also encourage you to look at Quest vWorkspace and the EOP extensions for RDP. In and on itself, RDP is lackluster, but EOP extends it in ways that are very similar to HDX. These extensions include JPEG acceleration, MMR, and many other upcoming enhancements such as Flash redirection.


@Peter, Interesting you left Quest, and that concerns me, because unclear what Quest will do without vision at the top. Nobody is providing that anymore :-(

What you call environment virtualization is fine until you get to Application Compat and the ability for any or these vendors to keep up with patching and certifcation, support etc. Ringcube is very cool, but app compat will be a concern, Parallels, can't patch fast enough. This is a MS problem to adopt in many ways before the industry will buy into it. If MS did the virtuozzo type thing for standard Windows server, then massive horizontal scaling benefits. However they seem to have chossen the path of just making TS better and gone down a different patch. So a consumer of the technology I have no confidence inrunning business apps on this. I have zero hope for this type of feature for Windows 8. I hope I am wrong and I mean that sincerly. I expect layers technologies that let apps behave as natively as posssible on the OS to be the only ones that will gain traction in this space. VMware won't do this because they want to virtualize everything, which will bring is back to app compat, they don't get the rest of it and will no invest in Desktop from what I can see.

Yes i've looked at Quest AKA provision networks stuff. Bottom line is that it's impossible to compete with Citrix so entrenched everywhere, and they match on price when it comes to it. Your protocol stuff is not even close to HDX yet in many ways but granted gettin better. The management of multiple hypervisors is good, but Citrix play that game too VMware doesn't.....

Even Quest suffers the sam concerns of a brokers. Unclear to me if the thing scales, because nobody I know uses it. My gut with Quest is that like VMware, this is a tier two offering with investment levels to match, it's not their core. Citrix has this front and center, and will outgun Quest, and get the space and care more than VMware. I wouldn't be surprised is Quest sold their desktop unit off to somebody. Perhaps VMware should buy them....

I agree


@appdetective, when are you finally going to come out? What agency do you work for? :)))

I think app compatibility issues will continue to exist to some extent, until Microsoft decides to embrace those types of technologies and begin to show some leadership and vision. Everything I've seen so far from them falls in the "me too" category. Of course, I'm referring to the likes of Ringcube, Parallels, and others. Even App-V can't promise 100 percent compatibility, and so no one particular technology is a panacea. The important thing is that choices do exist.

You'd be surprised as to how many organizations out there are embracing Parallels Virtuozzo Containers for their VDI deployments; I'd be the first to say that app compatibility is extremely favorable, and as far as patching, Parallels is doing a pretty good job keeping up with Microsoft. I think the story will eventually get even better once Microsoft publishes an API to be consumed by Parallels en lieu of the several hundred kernel patches that they have to put in place to do their magic. This will actually happen soon, not to mention that Microsoft will eventually come out and endorse OS partitioning.

I know of several large and extremely successful VDI implementatiuons employing both Parallels Virtuozzo Containers and Quest vWorkspace. Moreover, the vWorkspace broker is highly scalable, and that's based on feedback from very high-profile customers. If you think about it, "containers" are a more natural fit for mainstream VDI than traditional hypervisors: 1) Higher user density per server, 2) Inherent copy-on-write for storage footprint reduction...  

Yes, Citrix HDX is still ahead of EOP, but I don't think you've seen the whole gamut of features. For example, Flash acceleration has already been developed, but not released yet. And as far as image acceleration, it's gotten infinitely better in the last release, not to mention that the developement of progressive image display, as well as queuing and tossing, is underway. In a nutshell, the difference in RDP performance with and without EOP is night and day.

Having been part of the Quest organization, I assure you that virtualization is at the core of the company's strategy. This is especially true of the desktop space. Do not forget that Quest also owns ScriptLogic, one of the most prominent players in desktop configuration management.

Finally, you'd be surprised as to how much staying power Quest vWorkspace has, especially in light of the strategic relationship with Microsoft that Paul and I forged prior to our exit from the company. I don't have a bone to pick with the other competing vendors anymore, but I know the likes of Patrick Rouse kick VMware and Citrix in the pants on a daily basis.

Anyway, you articulate your points very well, so keep up the good work!


Great, it's more interesting to me to see containers on a Desktop OS. Only then can it be really VDI that solves many of OS specific issues like licensing. I don't see parallels in that space, and frankly still don't buy their stuff works based on when I tested it last. However i will keep an open mind. Wish you well in your new venture, and personally I hope Quest stays in this space as a balance to Citrix, because VMware will get killed. Hey Quest can I buy EOP without a broker it it's that good? EOP Connect....


VDI provides tangible benefits when a clean machine provisioned from a dedicated pool can also preserve the user’s personal computing experience without overburdening servers hosting virtual machines in the data center. Our approach can be considered persistency on-demand, which blurs the hard line between persistent and non-persistent desktops, through agile user virtualization.

The challenge then becomes how to best handle the user layer application installed on top of the golden image: settings, data, desktop look and feel and the changes that individual users make to the standard image VM.  Not all users need full persistency, but most require some.  Thus, to allow the maximum user experience, a more flexible policy-based approach is needed (similar to identity and access management) by controlling, through intersecting matrices, which user personality and application elements should be preserved between VM login sessions.  This includes the ability to support applications that are installed by a user on a virtual desktop but not part of the standard VM image.  Our vision is to achieve true single golden-image management, whereby from one standard image, IT can create multiple, customized desktops (VM images) based on user groups and flexible policy rules.  

Traditional Microsoft roaming profiles and other solutions on the market can do some of this, such as maintaining settings for certain applications which write in specific user registry hives, preserving data in a predefined location such as My Documents, but this still limits the end user’s expected computing experience. Ultimately, user personalization preservation needs to be addressed cost effectively and flexibly if desktops/laptops are to be hosted on servers in the data center.