Reviewing this year’s activity in the virtual desktop space has been very exciting. We have seen releases from almost all of the major vendors, and companies are beginning to truly adopt virtual desktops as a part of their overall desktop initiatives. I had the opportunity to speak with Aaron Cockerill, Senior Director of Product Management for XenDesktop at Citrix, recapping 2010, and more interestingly, looking at what’s in store for virtual desktops in the future.
2010 Recap and Challenges
JJ: 2010 was a very exciting year for virtual desktop technology. It was supposed to be the year for major VDI adoption. How would you characterize 2010? What were some of the challenges that your customers encountered?
AC: 2010 was the year that VDI became real. Many of our customers had tactical deployments that included call centers, developers and training rooms. Most of these customers had these deployments running for over a year before moving them into production. Since the release of XenDesktop 4, it was in our strategy almost entirely to focus on improving end user experience. We feel that XD4 achieved the level of user performance customers were expecting, and with the release of XenDesktop 5, we add simplification of administration to the story.
In 2010, companies made the decision to make wholesale migrations from XP to Windows 7. In reviewing how they were going to do that, desktop virtualization came up frequently as the preferred method for many of their users.
We saw the obstacles for customers were the lack of information on how to deploy virtual desktops to their organization. Things like hardware configurations and which user group to address first. Citrix marketing and technical teams developed a lot of information that was released at Synergy in Berlin this year that would help guide Citrix partners to more easily educate and implement virtual desktops with their customers. A program called “Wow-to-How” is a collection of items such as reference architectures and describing which constituency in a company should deploy virtual desktops to.
JJ: There are still many clients that have very tactical implementations and still haven’t got a vision on how virtual desktops will be part of the environment. There are some visionaries that really see how to leverage and manage the different desktop models (physical and virtual). There are many that don’t. They have very monolithic approaches on handling the enterprise desktop. How should the market help them get out of their own way?
AC: XenDesktop 5 does help those types of people from a simplicity perspective to deploy and maintain virtual desktops. They have to have a tactical need to deploy desktop virtualization in the last couple of years because it has been something that you had to be a bit of a technologist to get it right. The customers you are referring to that yet have the vision to move in this direction are going to be apprehensive, and if they are presented with a complicated testing approach, and even with CIO level asking they are going to say “too hard, too complicated” whereas with XD5 you can get a POC set up if you have a hypervisor configured in about 15 minutes. It cuts down the complexity barrier considerably and will help these customers out a lot. There are those companies that are not willing to take risk; they are going to be dragged less to a desktop virtualization strategy but more to a “corporate desktop as a service” model by their end users.
Speaking to clients’ CIOs, VPs of IT in Asia Pacific, Europe, and the North America have acknowledged that the retention of young talented people is now dependent on how they interact with the corporations’ IT departments. If these people want to connect using Macs, IPads or any tablet device and work from the café instead of the office and if companies start setting up hoteling offices to reduce the cost of real estate, that is just going to drag IT into a corporate desktop-as-a-service model. Whether they planned to or not. It’s not practical to do it any other way.
JJ: Application virtualization and packaging has been a key messaging for vendors like Citrix and Microsoft for some time now and is even more prevalent when talking about desktop virtualization because it gives the organization the ability to be dynamic and flexible with the new platform and it forces the customer to evaluate the application inventory they have. We are seeing clients who, even with a strong application packaging (MSI) practice, are going back and re-packaging in one of the application virtualization solutions to give them a better delivery model. The traditional ESD solutions are not working for them in this new space.
Citrix has always had a great place in the marketplace because they have always gotten how to “do apps” and how to deploy them. One of the challenges we have faced in the past two decades in the field is addressing application compatibility with Citrix and Terminal Services specifically. We are seeing that again with Windows 7. Today there are only a few application compatibility tools in the market. What do you think about why these tools have not been much more prevalent in corporate environments?
AC: I am completely familiar with all the available tools, but when Microsoft bought SoftGrid and included it in the Microsoft Desktop Optimization Pack , that really cut out other people from getting into the application virtualization market. Because you can’t really charge that much for it anymore if it is part of the same desktop license that companies are already paying for. There is an opportunity for people to do compatibility testing, but not many that offer compatibility solutions like Citrix streaming and Microsoft AppV. You can put customer’s applications into three categories: ISV applications, Legacy applications (ISV provided app where customer may be behind on support or release version or the ISV company is out of business) and Home-grown applications.
Since Windows Vista was released 3-4 years ago. Many of the ISVs worked to provide Vista support, which is very close to Windows 7. Users got Windows 7 compatibility easier than most major upgrades because of Vista.
Home-grown applications are still a major issue. And solutions from Citrix and application streaming from Microsoft AppV are the predominant solution to address these applications. Depends on what the home grown apps run on, but very similar to legacy applications, the customer may already be running them on some app virtualization solution, like XenApp.
Cloud Based Desktop Services
JJ: Citrix has signed up a large number of service providers that deliver a virtual desktop service. We are also seeing a strong push from Citrix to hosting applications and desktops in the cloud. I still see the same problems as we did over a decade ago with the application service providers (ASPs). Someone else owns the real estate, but the data being portable, secure and making it available back on the corporate environment seems to be the biggest challenge.
AC: I agree with you. But I will take it even a little further. I think we will get to a point where desktops and applications will be hosted in the cloud. Many applications are already hosted in the cloud, i.e., Salesforce.com, and are really successful. I think this will happen to a lot of other software. I think that desktops will also make a migration to the cloud, but I think there are a few things that are hindering that move; one is data. Until corporations and the legal infrastructure of many countries get to a point where ownership and management of information can be clearly controlled and defined legally in the cloud, I think there will be an apprehension on part of many companies to use cloud-based services. I also think what is hampering adoption is the federation of user identities which is necessary to make the user experience adequate for cloud based desktops and applications. Citrix is taking steps with their OpenCloud initiative to make federation of identifies a viable solution because of the single sign-on capabilities, but will remain to be a challenge until companies adopt that as a way of authenticating users.
The last challenge to make cloud services make sense is we need to take advantage of cloud economics. The challenge of taking advantage of cloud economics is a requirement that the service provider is leveraging multi-tenancy. Most companies I speak to today are apprehensive about sharing infrastructure with competitors or others in their space. These are the major challenges above software licensing itself. I wholeheartedly believe that it will be resolved, because the economics of leveraging cloud based services is so compelling and there are companies who have been able to demonstrate it effectively. I know we are going after the ASP model of the 90’s but I think we are in a good position to make it successful this time.
JJ: One of the things I found at the end of the ASP rage, was this onsite ASP model where we would still leverage the customer’s hardware; but instead of selling them the full boat licenses we would rent them the per user, per app, but everything on premise. They got the benefit of a managed solution without all the concerns of data and multi-tenancy. Also the performance of LAN/WAN and Internet at that time was questionable.
I believe that we will see mobile wireless speeds in this decade get so much better that it will be almost the same as our corporate wireless. Do you think that we will get to a point of consumerization of virtual desktops, powered by the Internet and wireless service providers? Or will there be a time when people will be at home and they will order a Dell PC and will actually get a slim-line device and get a desktop service provided by Dell instead of a local installed OS?
AC: I believe yes, without a specific timeframe. I expect that it will be available in 3-5 years and will be a viable option for computing in 5-8 years. We are already having service providers and hardware manufacturers talk to us about providing services like that. You can see from everything from the Mesh capability in Windows and Apple’s MobileMe where consumers are rapidly adopting the cloud and the apprehension about data has already been overcome, or largely overcome.
I think the consumerization and the rapid consumer adoption of cloud based services will drive desktops to be provided in the cloud. The only thing I think will be contrary with that is that when we start to develop a better way to interact with computers, such as the touch capability on iPhone or gestures on Kinect for Xbox360. In the same timeframe there is going to be more natural interaction from a user perspective, but more complicated from a computing perspective methods of interacting with desktops and applications. Everything from speech to gestures to touch and the ability for virtual desktops being delivered especially in the consumer space to be able to be able to deal with those methods of interaction will be required if virtual desktops in the consumer space is going to be successful.
JJ: So do you think voice commands will take over touch?
AC: I don’t think that voice will take over touch. The adoption of voice is quite slow. The original implementations (Dragon for example) gave users a poor experience and it has gotten a bad rap for it. Majority of development of voice recognition on phones has been poor even though you can rely on it on phone devices and in your car; the adoption in other areas has been very slow. Whereas touch has been so beautifully executed by Apple where it is being adopted very fast. Voice and gestures will become an increasing big part of the way we interact with computers, and must be integrated with the virtual desktop experience.
Influencing emerging technology
JJ: We have seen over time that adoption of technology in the corporate space is many times driven by end users. I can recall companies investing heavily into ISDN solutions in the 90’sas a WAN solution because it was labeled as the best new home technology in PC Magazine. Even Citrix keynotes showcase consumer based devices heavily. What consumer technology is out there waiting to affect corporate IT?
AC: Definitely the tablet devices are going to be the killer one, but we are not limited to it. I think it ranges from cloud-based services for storing personal data to the devices that are coming to the market. The way the user interacts with everything from email to data to the devices they use to the way they keep everything in sync is going to be an expectation of corporate IT. The fact that most corporations are still struggling with roaming profiles and synchronized data working across sites whereas MobileMe and Mesh in the consumer space already have that solution solved. So I think the tablets and the sync capability is going to be expected.
JJ: Do you think we will ever see desktops on the Wii and Xbox?
AC: Only if the integration of gestures works. There is no longer going to be a distinction between what is hanging in your lounge room being a computer or being a television. I think it will be a big screen computer. I think people will look back and ask why did we ever have the two? The distinction will go away.
Profile Management in the Cloud
JJ: Let me talk about profile management for a minute. AppSense has a vision of support of full federation and portable user profiles as a cloud service. What are your impressions of this sort of vision?
AC: I fully support them going in that direction I think that all of us need to go in that direction to enable the success of the cloud.
JJ: Are Citrix and AppSense working closely together to make that happen, or separate efforts?
AC: It is definitely a shared vision and not a joint development effort.
JJ: The part that I cannot wrap my head around in AppSense’s or any product’s ability to control and move around all the user data in a cloud service. There is so much data in the user profile. Do you think this is also a challenge, I mean the profile size issue?
AC: Profiles can get out of control. And a big part of that is to choose what does and does not end up in the profile, for example Java app caches. The space that is required for typical profiles is going to increase because of the use of more and more multimedia. It’s not like you use all of that data at once and have to pull down all that data to the machine you are on. Our vision around profile delivery is that you only ever deliver to the OS the references for the information if it even needs it, and you only pull down to the relevant device the pieces that are going to be, or can be, used. If you can deliver the pieces of the profile on demand, then I don’t really see a problem from a technology or size perspective residing in the cloud. The challenge is about data ownership. If your vision of connectivity speeds does come true then I believe where the connectivity is indeterminate from corporate LANs and WANs and wireless, then the delivery of an on-demand profile is going to be trivial.
JJ: Citrix has an approach very similar to Microsoft regarding XenDesktop licensing, which is based on named user or device. For the large customers who have leveraged concurrent usage (CCU) licenses and maybe have a “follow the sun” model, the trade-up conversions of 2:1 just don’t make sense. Will Citrix look at the trade up programs again in 2011?
AC: Admittedly I am not part of the sales organization and don’t have too much insight on the future of licensing. We are choosing a similar model to Microsoft because all of our products are dependent on or linked to use of a Microsoft product. Desktop virtualization is all about virtualizing a Windows desktop operating system or a terminal server CAL. The alignment is not just from a partnership perspective, but because it makes our customers lives easier. Customers might dislike Microsoft license models and wish they were provided differently. If Microsoft were to change their models again we should rapidly adopt a similar approach. It makes them easier if they can count licenses the same way for the license and the solution delivering the license. If the Microsoft licensing is annoying to them, they should go back to Microsoft and let them know how they can do it better. I don’t know of a situation where a customer says they love software licensing.
What kind of guidance would you give customers to move forward in 2011?
AC: The most important thing is to pick a solution that will give you the ability to choose later, or change your decision later. The other thing is to review the Wow-to-How from Citrix with your Citrix partner. It makes the consumption of desktop virtualization tangible to most IT folks. It also does it in an approach that makes the IT folks look like heroes instead of being the people who got dragged into this by users. IT folks can benefit from the thought process that has gone into developing this program. It was developed by interviewing and understanding current deployments and getting an understanding of the financial ROI and delights of end users perspective. Companies should do whatever flavor of virtualization that is appropriate, for the right user constituent, in order to have the overall strategy that is successful. What needs to be done is categorize your users to clearly understand their needs. Line them up with the right technology, but then roll it out where you can get the biggest bang for your buck. So the users you roll it out for first will tell other people in the organization. The program really helps IT to become the winners in this situation. They are lowering costs and making people happy. They can accomplish this by hitting the right groups first with the right technology.
2010 has definitely proven to be the year that virtual desktops have become a reality. Similar to what we had seen with server virtualization in 2005, we are now coming out of the cloud of fear about this new technology. The time of kicking tires is over – get into the car and lets go.
Top things companies should be looking at in 2011:
- Application packaging, virtualization and compatibility in virtual desktop environments – populate the service catalog with these offerings
- Really engage with your users. Understand their needs, how they work, and how they want to get to their apps and data.
- Think differently about desktop management. The life cycle of a desktop has changed, so have the people, processes, and tools to manage the various desktop models.
- Take a look at your cost and charge-back models and embrace a cloud economics approach where the user will get more bang for their buck with greater flexibility and performance.
Thanks to Aaron for taking the time to speak with me and providing excellent insight into his vision of desktop virtualization futures.