 Okay, so my name is Ken Ringdall, Vice President of Engineering at Destone. We are a software company. Give you a quick agenda here. So here to talk about cloud hosted desktops and bringing cloud hosted desktops to a cloud platform like OpenStack. So what I'm going to do is I'll do a quick introduction to Destone, who we are, what we do, give everyone a little background there. Talk about the evolution of the desktop. The desktop is of course ever changing, but there's been significant innovation to the desktop in the last five years specifically, so we'll go through that. Understand what desktops in the cloud really means, who's it good for, what are some sample use cases, what are some of the impacts of doing desktops in the cloud. Talk about the market opportunity for desktops in the cloud, virtual desktops and desktops in the cloud. Give you a little bit of an architectural overview, technical audience here, want to give you folks a little bit of an understanding of how a DAS platform works, specifically how we're architected, how it's different from enterprise VDI. When you bring cloud hosted desktops and OpenStack together, go over some of the benefits of bringing those two together. And then I'll leave plenty of time at the end for questions. And I'll leave most of the questions for the end. If anyone has a burning question, they've got to just raise your hands. We can certainly stop, but I'd prefer to move the questions to the end. Before I get started, just a quick show of hands. Who knows what VDI is, who's using VDI? Who here uses a virtual desktop on a daily basis? It's your regos. It's good. It's more than I thought, actually, which is good. Okay, so who is Destone? So we're a software company. We were founded in 2007. I was on the founding engineering team. So I've been at Destone since we raised our first round of funding. You see a sample of our service writers on the left here and a sample of our customers on the right. So we are a virtual desktop platform that's built for our service providers. So from day one, we built a software platform. We have our own intellectual property. We don't leverage any third party broker technology. The technology we have was built on our own. When we set out five, six years ago and found a Destone, we looked at the market and in a few moments here, I'll go over kind of the evolution here. But we determined that we essentially founded what's called desktop services service, which is really a service provider delivering a desktop service, which is different from your internal IT organization, even though some IT organizations consider themselves service writers to the enterprise. But we had a specific focus on service providers delivering a hosted solution as a subscription service. We have eight patents on desktop as a service. I have a slide that goes through that in a bit. But we are headquartered just outside Boston, Massachusetts in a town called Lexington. It's just a suburb of Boston. So desktops are changing. Of course, in the 1990s, we had Windows 95, Windows 98, really the first really client operating system in the desktop, Windows operating system. Lots of legacy apps that were in Windows, but the desktop is changing. So as we kind of look at it, 2008 we look at it as really the evolution of VDI. There's been lots of desktop virtualization prior to that. But we really look at that as kind of the beginning of the change to a virtualized desktop and moving a desktop into the data center. You see the evolution of operating systems there. But it's changing. So you bring your own device. How many people have iPads and Android tablets and want to connect to their desktop and don't want a full desktop session? Maybe you just need an app or a file off your desktop, and you don't want to have to be at your desktop. So bring your own devices is radically changing the landscape of desktops. Everyone's got their iPhone devices and their Android phones and iPads that they're bringing into the enterprise. And the enterprise is having to adapt to that. And virtual desktops really helps the mobility use case there. The migration, right? Everyone knows Windows XP is going away. Microsoft has been warning people for years and years. But there's still a lot of Windows 7 migrations coming. Windows 8 is coming in a week and a half or so here. Mobile employees, right? Lots of large companies are transitioning to work at home and smaller branch offices and sending people out of their office because it's expensive to keep the lights on and pay for heating and all that expenses in the office. So they're increasingly sending people home. And what do you do for people that are at home? You send them with a laptop. Well, when their laptop breaks and they're at home, they're dropping it in FedEx overnight. And what do they do for the 24-hour turnaround it takes to get another desktop back? So the desktop is changing. And there's also security and IP concerns. I mean, every couple months you hear a story about a laptop that had some kind of special file on it and got stolen and had social security numbers or some kind of passwords on it. So there's security and IP concerns as well around the desktop. And how do you make the desktop more secure? But so people have moved to virtual desktops. But virtual VDI is hard. As I mentioned earlier, when we found a desk town, we knew that the VDI evolution was coming. And people would struggle. So VDI is very different from running servers, lots of upfront CapEx costs. So instead of paying for physical PCs that go into people's desks and you can replace them on a one by one basis or refresh them in small cycles, instead you're buying large infrastructure, network storage, compute capacity in the data center. You're having to allocate more data center space. And we have lots of customers that come to us that say, we can't do VDI on-prem because we just don't have any power left. We don't have any power. We don't have any space. We can't do it. We need someone to host this for us. So those challenges are there. Lots of expertise. People that go do VDI, they have to hire full-time staff to manage the VDI environment. They need experts. Sometimes these enterprises are not large enterprises. They need to get a SAN administrator or some other specialized talent to manage the VDI environment. It's complex. Anyone here who's done VDI knows that it's not just about layering a hypervisor and a broker on top of it and you're done. You've got your Windows virtualized operating systems. Anyone who's done that knows that it's all about the user experience and the protocol and all the other things that go with it. If you've looked at advanced VDI, maybe you're looking at layering applications and having the app stream in and profile virtualization. So it's hard. And there's lots of learning to do. And when there's lots of learning to do, either hire expensive consultants or you learn on the fly and you make mistakes and have to adjust. So doing on-prem VDI is hard. And that's the business challenge we looked to solve was let's take that challenge away from them and put it in the hands of a service rider that does this at scale and can help the enterprise do that. And of course performance. So performance is incredibly important to a desktop experience and anyone that any customers we talk to, the first thing is, is it as good as my physical desktop? Can't do my apps run? Will I experience any latency or will I see any slowdown in my desktop? Very important that performance of the desktop is just as good and that comes down to fine-tuning the LAN or WAN connection to your desktop, fine-tuning the desktop itself. Anyone who's tried to take a regular Windows image and virtualize it and there was a whole concept of P2V, it's challenging because there's lots of optimizations. Running a virtual instance of Windows is very different. You have to optimize. You can't do regular security scans. There's other things that you need to tweak in the registry. So it's challenging and that's why it's a great opportunity for service riders to help solve that problem for enterprises. So our vision, desktops in the cloud. We saw 1990s desktops in a PC, physical device. 2008 is about when we said desktop in the data center. 2010 is when we kind of look at as really the beginning of desktops in the cloud. You can leverage it as a service model. So instead of putting all the capital expenditures up front, the service provider who does that for a business, they're used to buying capital equipment up front and burning that down and delivering a subscription service. So you can leverage as a service model, move capital expenditures to operational expenditures, centralize management. So moving your desktops, and this is part of EDI as well, but centralizing your desktops, moving your desktops and virtualizing your desktops is supposed to help the desktop management case. Instead of having your desktops spread across the enterprise, your desktops are now local in the data center. You can control them there in one place. It's supposed to help things like patch management and other things that are expensive processes to manage physical desktops. Data center proximity is a very key thing because when you think about taking your desktop and now moving it to a hosted data center, the question is now how do I get at my apps and my active directory and other things that are at my enterprise? We call this the backhaul. So basically when you move your desktops to the cloud, those desktops logically live on your network. Your own enterprise at IP addressing, you can reach those desktops from your corporate LAN, and those desktops can reach your applications. They can reach your exchange. They can reach your SharePoint and everything else that you have on the network. And that's because there's a site-to-site connection there that's either, in MPLS link, in some cases it's a site-to-site VPN. But basically we allow the ability for those desktops to reach back, and they're logically on your network. So to a user, they appear like nothing's changed. My desktop is still there, but it's now hosted in this service provider data center. And then of course the cloud, everything's about elasticity and scalability. So moving to the cloud is all about flexibility and being able to scale up and scale down. OK, so when you move your desktops to the cloud, the question then becomes, OK, what do I do as the enterprise and the desktop administrator, and what does the service provider do? Above the waterline here is managed by the client. So the client still does their desktop management. So if they were patching their desktops and they were managing their desktops, supporting their desktops, they still do that as they do today. They can access them just like they're local. Access devices, so endpoint devices, whether they're repurposing physical PCs that were retired, whether they're purchasing thin or zero clients, that's still owned and managed by the client. Applications, all the applications that sit inside the image and other hosted applications that they have back at the enterprise, they still own that process. They're in the desktop, they're owning that process. The image creation, very important part of doing virtual desktops is getting that image fine tuned, making sure the correct apps are in there, the right settings are in there, you're connected to your active directory and you have everything you need in there, still owned and managed by the enterprise. And you're operating system license. So a very question that comes up very often with doing virtual desktops is the Windows licensing. And specifically when it's hosted by a service rider, it's even asked, can I do that? Can I buy my Windows license from the service rider? Well, the answer is in some cases yes, in some cases you need to bring the license to bear. So I won't dig deep into Windows licensing but in short, if you're running a Windows server, for example, a service rider can provide a splot to you and you can rent that desktop on a monthly basis and pay for it like a subscription like you do, any other service you get from your service rider. A Windows client OS license, Windows XP, Windows 7, Windows 8, not quite the same, some of the licensing restrictions are different. The client brings that license to bear, some service riders have relationships with the right distributors where they can actually provide what's called a VDA license to the customer. It's in the customer's name but they can provide that as bundled as part of the service. But if an enterprise has software assurance, that's covered as well. So that's all managed by the client. Down below the water line here is everything else the service rider now handles. So when you move your desktops to the cloud, these are the things that you don't have to worry about. We talked about data center facilities, you don't have to worry about power, cooling, rack space, anything else. The facilities are taken care of you by the service rider. Software, so not software that's in the desktop but the virtualization software all handled by the service rider. So that is the hypervisor software and that's a platform like the desktop platform that is providing you the brokering and the management of your virtual desktop all handled by the service rider. Upgrades all handled by the service rider. You don't have to worry about moving from one version to another one. That's done for you by the service rider. They do it in the maintenance window at a time that's convenient for you. On the hardware side, we talked about moving capital expenditures to operational expenditures, right? No longer are you having to shell out, just to give you an idea of the cost of EDI. I mean just to do say a thousand virtual desktops you're probably talking about a half a million dollars worth of capital expenditure. So now an enterprise is not having to shell out all that cash up front. They move that to an operational expenditure and pay their service rider for that. And then support. There's always so many, a service rider has SLAs that they deliver. They provide the 24 by seven support. Their job is to make sure your desktop stays up and running all the time. Okay. Give you an idea of the market. So, this is the total market size estimate. So the big blue there is physical PCs. So of course in 2010, that's a very large portion of the market. The blue line that starts turning up down the bottom and starts turning up, that is VDI. And that's all of VDI. So that can be hosted session desktops. That can be on-prem VDI. In 2014, is number I'll call out with the little bubble there. 2014, 20% of desktops, virtualized desktops will be hosted. That equates to about a $3 billion market. So a really large market. And one of the things that, we're here to talk to the OpenStack community is about bringing a brand new market to the OpenStack community. And really talk about some of the challenges that that brings. And I'll get into that in a future slide here. But a really large market opportunity. I mean, everyone knows about the tablet, the tablet wave here. You can see the big green block there in 2014. But clearly virtual desktops are on the rise. And the physical desktop is on the decline. So the desktop is clearly changing. Okay. So a sample of some of the folks that are delivering a hosted desktop with the desktop platform. So four different types of folks. Top left, Dell has a service called Simplified Dads. We look at Dell as assistant integrator. So they're a services provider that can provide you lots of services. Hosted desktop offering is one of them. A managed service provider. So these are folks, typically they end up being sort of regional providers, but they're one-stop shopping so they can provide desktop support. They can provide patching of your OS. They're basically, SMBs love these guys because they say, look, I don't want to deal with my desktops, just do it for me. And they do everything. They can host the desktop with a DAZ platform and they can do all the management of the desktop for you. Support, patching, everything. Telco providers, so BT is an example here. But Telcos have big pipes. They've got infrastructure. They're already delivering some cloud services. Doing DAZ is a logical opportunity for them and we see lots of Telcos expressing interest in lots of them providing desktop services. And then another partner of desktop called NaviSite, NaviSite's a cloud provider. So they're already providing infrastructure as a service. They've got a large investment in infrastructure. Layering a desktop service on top of that makes total sense for cloud providers. So it's just another cloud service. One might ask, how is doing DAZ different from a cloud platform? So we get asked the question a lot is why can't Amazon do what you're doing? Why can't a cloud platform do what you're doing? It really comes down to the three things here, which is session management. So cloud platform has no concept of a session. So what a DAZ platform or what a VDI platform does is it knows when people are logged into their desktop, when they're disconnected from the desktop, when they were allocated to the desktop provisioning. So provisioning of desktops and managing large-scale desktop pools is something that cloud platforms and cloud services, infrastructure services, typically don't do. So you want to put your engineers in a specific group pool of desktops because they have a different profile and they have different apps and they have different services that you provide to that group versus your sales group versus your marketing group. So a DAZ platform and the desktop platform provides that notion of pooling and being able to group people into alike groups. And then I talked about session management brokering. So being able to do like a pooling concept where you over-allocate or over-subscribe users. So you may have shift workers or seasonal workers or we're having a one-to-one desktop. It doesn't make sense. It's more costly and it's overkill for what you need. So if you have three shifts a day in a call center, you don't need a desktop for every one of those. You can rotate that. So you can do, for example, a non-persistent pool and over-subscribe that pool and the desktops will, you know, someone logs out of that desktop that will revert, roll back into the pool and be ready for the next person. An example of some things that the desktop platform and a DAZ platform do that Cloud doesn't do today. I talked about patents there. So some of the patents that we have on our technology. I mentioned our technology is very different from Enterprise Class VDI. So Enterprise Class VDI is very tuned and very designed for serving desktops to the Enterprise single-tenant on the LAN. The desktop platform and a DAZ platform is architected and built differently. It's really designed for multi-tenancy, just like a Cloud platform is, and delivering it as a hosted solution over the WAN. So everything we've done has been to optimize over the WAN and optimize for multi-tenancy and provide that management built into the platform. So very different use cases from Enterprise VDI. OK, talk a little bit about the DAZ platform. I'd like to show this global view of the world because when you think of a DAZ platform, multiple data center support is very important to that service. So if you're a large enterprise, you have users say in both coasts of the US, in Europe, in Asia-Pac, you want to manage those users the same way as regardless of what geography those people in. And if I fly from the east coast to the west coast, I don't want to have to know I've got to point to a different broker and I've got to manage those desktops differently. It's all handled under the cover. So multiple data center and multiple geography support is built in. Again, it's another difference between Enterprise VDI and hosted DAZ desktops in the cloud. Up top, I'll call you attention to the different desktop types. So I talked about regular Windows client OS, RDS, or session-based desktops. Very popular for a certain class of users. The desktop platform can do RDS. It can do Windows regular Windows desktops. It can do Linux desktops. So we see, especially in the developer community, and some of the folks that are here, a Linux-based virtual desktop is desirable for certain use cases. And we also do Linux-based virtual desktops as well. So lots of options provided there. We talked about the licensing of the desktop. Down the bottom, I'll talk about what's there. I mean, basically, the DAZ platform, we layer on top of virtual infrastructure. So the hypervisor and the compute network and storage, very important thing. What we provide to our service writer partners and our customers, not only the software, which is the core intellectual property of desktop, we also provide what we call a blueprint. And that blueprint really is a recipe or a prescription on how to do desktops as a service. It describes how to do multi-tenancy, how to configure your networking to use VRFs and layer to VLAN security to provide that multi-tenant solution. It describes how to configure your storage, what your needs are for your storage, what type of storage you should have. Talks about the compute and how our platform integrates with the hypervisor. Very important because it really gives the service writer a good idea of how they need to build their platform. Because desktops is not something that's called a service writer. They're typically most service writers are infrastructure providers. They don't know much about providing a desktop service. So our blueprint is a big piece to helping them get going and prescribing them how to deliver that service. Okay, so the evolution of Das. So I just talked about the fact that we layer and we talked directly to the hypervisor. We talked directly to storage. We talked directly to networking. Quite honestly, when we started in 2007, cloud platforms didn't exist. We had no choice but to talk directly to all of these systems individually. And if we wanted to support different devices, different storage providers, we had to build a plug-in for each one of these. So we spent a lot of time working on this virtual infrastructure layer. And so that was phase one, building on what we call physical infrastructure. Phase two is really building on top of a cloud platform, building on top of OpenStack. You know, many benefits there, abstraction from all those virtual infrastructure components, abstraction from the hypervisor, the compute, the storage, the networking. You know, it really provides a good opportunity for infrastructure partners where instead of engaging and having to work with specific APIs, it's more meet in the middle. So desktop engineering builds, we build our platform to stay the OpenStack APIs and then the vendor builds on the bottom side. So we meet in the middle and it's a good way to get common support for infrastructure there. And it's a great ecosystem for partners, right? So it's a great way for us to evolve. And when we go talk to service writers, they all have their choice of who they like to work with from an infrastructure perspective. From a storage perspective, some like NetApp, some have relationships with EMC, some have relationships with Dell. You know, quite honestly, you know, there's advantages and disadvantages to each, but service writers have their choice and we wanna give them their choice. We wanna be as accommodating as we can to our service writer partners. So it's all about giving that abstraction and really building an ecosystem that allows for that. So to give you an idea of, you know, how is this architected, right? So talked about being a multi-tenant platform. So you see here, this is showing a picture of two different tenants. They're running side-by-side. Each tenant gets their own broker. We integrate with, as we said, we have a backhaul connection to integrate with the enterprise. So that includes Active Directory. So we use the customer's Active Directory. This isn't like a shared solution where there's a shared AD that everyone connects to. It's the customer's AD. We integrate with it. Sometimes that AD is over the backhaul. Sometimes they'll stand up, a replica AD in the data center. So it's close by the desktops. Doesn't really matter to us. We'll integrate either way. But every tenant has their own brokers. You know, their desktops are logically living on their network. Different desktop types. The common layer down the bottom is where some of the core intellectual property comes in, the resource manager. I talked about phase one being that we ride on top of physical infrastructure. So that resource manager is a common layer that talks to the compute, the networking, the storage. From the tenant side, the tenant has no idea whether they're running on top, what hypervisor they're running on top of, what storage is on the back end, what networking is in the back end. Quite honestly, from a service writer perspective, it shouldn't matter. They're providing an SLA with a quality of service and they're delivering that to the customer. And that's what they're held to. And so it's all about delivering and letting the enterprise deal with their desktops and letting the service writer deal with the virtual infrastructure. So that's really what the picture looks like today. Compare and contrast that to what the picture could look like with OpenStack in the picture. So tenant A, tenant B is still there. They each have their own broker, desktops are logically living on their network. The difference here is OpenStack is now slid in the middle there, right? And there's no direct access from that resource manager component to the virtual and physical infrastructure. OpenStack is in there, right? We've got Nova on the compute side, Quantum on the networking side, Cinder, Glantz, Swift for templating and backing store for the virtual images. So that's what the picture looks like and that really gives us lots of flexibility and lets us go focus on what's core to us, which is providing a great desktop platform and not necessarily managing virtual infrastructure. Okay, when we engage customers and customers, not the service writer partners, there's real two things that they ask and two things that they want out of a desktop service. The first thing is everything comes down the cost. So they always ask, is it as cheap as my physical desktop? The answer sometimes there is challenging because for the most part enterprises don't have a good feel for what their cost of a desktop is. They know what they pay for a PC. They know what it costs to patch it. They don't know what it costs to really manage it and support it over time. So those estimates very widely, but it is about cost and really, is it as equal to our physical desktop? It's also about performance. Does it perform well? Will my users complain? Will they be happy with the desktop? Those are the two things. Moving to an open stack platform to tell the truth about this picture here. So you see percentages here. I took out some of the hard numbers. We've got a very, very complex model that we put together around the cost of virtual desktop. We know what power and cooling costs, what space in the data center costs. We know what storage costs, the compute and the hypervisor costs. We have a very detailed cost analysis. For the purposes of this I stripped those out, but this is an example of the savings that can be delivered by moving from a virtual infrastructure, building on top of virtual infrastructure to building on top of open stack with an open source hypervisor and other components to help drive down the cost. So clearly big savings on the desktop virtualization license side. Operationally, moving to a cloud platform, moving to open stack, really allows to drive down the cost. Today, to get a customer set up, it involves configuring storage and configuring network and configuring hypervisors. Moving to a cloud platform allows us to automate a lot of those processes. So for us, the Holy Grail is all about an online try and buy model where somebody goes to a website wants 100 desktops and a couple of hours later, they're connected to their desktops. And everything is magically done behind the scenes, not by a human being, but by software automation. And the components are there to go do that. So moving to the cloud platform really enables that. Doing okay in time here. So why open stack for DAZ? Desktone is a really strong proponent of open source. Our management platform leverages open source extensively. We have a Linux Ubuntu appliance. We have an open source application server. We have an open source database. We use the Spring Framework. Open source, mature, friendly open source is a good friend of ours. So same goes for open stack. We talked about costs, we talked about being vendor agnostic and clearly everyone's here this week understand that there's strong community support here. It's a very important piece of moving to any open source is really all about is there a good community support? So that's a big check here. So the value of open stack for DAZ, it allows us to get away from managing virtual infrastructure and really focus on a great user experience. And that's really, it's all about making end users happy. If end users are happy, the service will get adopted and that allows us to really focus more of our time on user experience. Scalability, being able to get to tens and hundreds of thousands of desktops. So the scale of a cloud platform really important to a DAZ platform. Elasticity, everyone knows when you go to a cloud, it's all about can I scale up, can I scale down. Good example here, we have customers that are seasonal workers. So at tax season we have customers that say look, I need an extra 500 desktops in tax season and then that's gonna go back down to zero. I don't wanna buy 500 desktops and pay for them for over three years or such. I need those 500 desktops for three months. Great use case, contractors, dev and test, good elasticity use cases. Vendor flexibility, we talked about open source, automation, shared infrastructure. Many providers we go into, we're not the only ones that are riding on top of their virtual infrastructure. They're running platform as a service, they're running databases as a service, they're running all other services and they say look, you need to come in and you need to play nice with everyone else. We can't dedicate large portions of our infrastructure just to the desktop service. So being able to share that and riding on top of a cloud really enables us to do that. So OpenStack will open those opportunities. And then working with the OpenStack community. So we're just getting engaged with the OpenStack community. Part of this talk is really about really opening people's eyes to the fact that desktop is a very different workload on top of, very different workload on top of a cloud platform. I'd say the large majority of the OpenStack use cases to date have definitely been infrastructure service. Das is very different. Talk about the storage problem. Storage is huge in virtual desktops. The random IO of virtual desktops, the high right activity, the scale 500 servers is a lot of servers. 500 desktops is like a drop in the ocean. So the scale is very different moving to bringing Das, desktop of service on top of a cloud platform. So we want to engage with the community but we also want to ask for other people's, other folks's help with that to work with us to help really move the platform along with some of the changes we think would be really helpful in general but are useful for the Das use case. Couple things that we want to do here. HA for desktops, very different from servers. Different use case, we've been sitting in some of the design sessions there. Future VDI is non-persistence. So I talked about layering and profile virtualization. The holy grail of virtual desktops is to get to a point where none of your data is sitting inside the virtual desktop itself. It's all layered in and separated. You can detach that all, you can lose your desktop and you have all your data and all your profile and settings, et cetera, very different use case from servers. We talked about the number of VMs. And Das encompasses all of the OpenStack projects. So Nova, Glantz, Swift, Cinder, we're going to engage with all of them. All of those components are key to the desktop service. So I'll stop there. And I think we've got a few minutes. Ask some questions and I'll stick around too if we run out of time. But first question. We are not using OpenStack today. What really this is about is, we're looking at the landscape and we're looking at OpenStack to really provide a lot of value here. So to mention, we're just engaging with the community. We wanted to talk about what Das would look like on top of OpenStack, but it's something that we're looking very closely at and stay tuned for some news in the future. We do, we do. We do, yeah, so just to, sorry, I should be repeating the questions here. So the question is around, do we mitigate the login and the boot storm? Common problem with VDI. The answer is yes. With persistent desktops, typically they're up and running, so we don't have a lot of reboots on those desktops, but we work very closely with our infrastructure partners. So part of the blueprint I talked about actually is a prescription on how to design your storage and design your infrastructure to really be resilient to those problems. They are real problems in a VDI environment. For user data? Yeah. Yeah, so the question is, well, do people wanna use their own storage for user data? And absolutely, yeah, so that's the back call connection I was talking about. So absolutely, the desktops can reach out to that storage and use that, absolutely. You're talking about multi-tenancy, so if you go in at a client and they've got all their Windows licenses, do you force them to throw all those Windows licenses away in order to go to the multi-tenancy licensing that Microsoft requires you to have? No, so they can leverage their existing licenses. It depends what they have. If they have software assurance, no problem. If they have VDA licenses for their existing Windows, no problem. If they're adding on to what they already have, they need to buy licenses for what's there, but we leverage exactly what's there. The multi-tenancy, the problem, and not to go too deep on the licensing because it becomes a little rattle, but the challenge, Microsoft has a specific rule that says that if a customer's running a Windows client OS, it needs to be on non-shared hardware. So customer A and customer B have to be on physical, separate compute. Our platform ensures that that happens, and we've ensured in the early research we've done with the OpenStack platform that OpenStack can do that for us as well if we orchestrate it properly. Okay, so you're putting in rules to keep them physically separated. Absolutely. But you're still calling a multi-tenancy. It's multi-tenancy because the service provider is delivering a multi-tenant solution. They're on separate physical hardware. Now in a Windows server desktop or a Linux desktop, they absolutely can be on shared infrastructure and we support that. Okay. Next question? Do you actually have a roadmap or a timeframe for completion of phase two? We do. So we do. It's not 100% public at the moment, but I'll kind of just put it out there in a broad spectrum and say 2013 for sure. Other questions? Okay, I'll stick around for a few minutes if people have other questions, but I appreciate the attention. Thanks very much.