"Circling the drain" is the phrase I recently used to describe Desktone to a friend. The company's been around for awhile with a vision of cloud-based hosted VDI desktops. They made a lot of noise a few years ago but I hadn't heard much of anything about them more recently.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
But last year, Desktone announced a new strategy. Rather than trying to sell their software to huge providers (IBM, Verizon, etc.) who would then in turn offer hosted desktops to their customers, Desktone decided to make a bold bet and do it themselves. They bought a bunch of hardware and built-out their own environment in a Rackspace colocation facility, and now they offer Windows 7 Enterprise desktops in the cloud, direct to the customer, for as little as $1 per day.
Desktone struggles through 2009 & into 2010
Prior to last week, the last real conversation I had with anyone at Desktone was in April 2009. (Shame on me? Shame on them? Dunno.) At the time they were a software company building a software product that could massively scale desktops running in a datacenter but tuned for service provider needs like multitenancy, distributed administration, image management, etc.
That's an important fact which I missed the first time around. Desktone was a software company, not a services company. Their software product actually competed with Citrix XenDesktop and VMware View in the eyes of the service provider. (What I mean is, if you're a service provider who wants to build a cloud desktop offering for hundreds of thousands of users, you need to decide whose VDI product you use: Citrix, VMware, Desktone...)
Desktone felt that since they were only building their product for massive scale which was only targeted towards server providers, they'd achieve success through their deals to these huge providers. And the providers would choose them because they'd be getting a purpose-built VDI solution that's ready to go out-of-the-box. They wouldn't have to try to take something like XenDesktop or View--which works fine as a single-server ten-user solution--and also scale it up to 100k-user grid.
Early customers included HP, SoftBank, Marubeni, Verizon, IBM -- all with the idea that they'd use Desktone's VDI software to power their own cloud-based Desktop-as-a-Service (DaaS) offerings.
The problem, however, was that there wasn't a lot of traction.
Desktone would later realize that a big drawback of their model was the sluggish timing of their approach. It would take 6-9 months right off the bat for the telco to really understand Desktone and to figure out how they'd do it. Then it would take another 6-9 months for the telco to build out their solution, followed by another 6-9 months for the telco's sales force to learn how to sell it. Add that together and you've got an 18-24 month cycle from Day One until the telco is really selling. And in this world, that's an eternity.
Of course there are other downsides to only selling through an intermediary. In the telco model, Desktone is removed from the actual end-customer feedback loop. Do customers like it? Are they using it? What features are important? What needs to be improved? Any answers to these questions come from the telco themselves, so it's hard to differentiate between what the telco wants and what the paying customers want. (Although I guess both are important. :)
Finally, this "only available through enterprise providers" model meant that Desktone couldn't answer simple questions like, "How much does it cost?" or "Where can I sign up?" (This is one of the main reasons I thought they were shady before. I wanted to try it, but they didn't have a direct way to do that.)
A new strategy for 2011+: Desktone Direct
In mid-2010, Desktone ended up with a new CEO, Peter McKay, who also brought over a few of his pals from years back. With the new CEO came a new strategy: Desktone would use the software that 25 developers spent four years writing to build their own cloud which they would use to sell Windows desktops as a service directly to end users. They literally bought racks and racks of hardware (from Dell, NetApp, & Xiotech) and installed it in a Rackspace colocation facility, creating their own cloud datacenter.
The new direct model gave them a lot of advantages:
First is that Desktone can now be the master of their own destiny. They can build their products, price them the way they want, and market and sell them the way they want. They're closer to the customers for feedback, so the products can get really good really fast. And they can incorporate new features quickly. This has led to them creating several ancillary offerings, like domain joining capabilities, VPNs back to on-premise servers, cloud-based file storage and network shares, and master disk image versioning & management.
Another benefit is that pricing is very straightforward and Desktone can customize the offerings as needed. (Need 100 desktops but with 4GB RAM instead of 2? No problem. That can be ready in just a few minutes.)
Finally, the direct model means that Desktone can more easily sign-on partners. MSPs who were offering complete desktop management to their customers could also now offer "the complete desktop," thanks to Desktone.
(Oh, and I guess I'll just say now that I think the name "Desktone" is a really clever name, so I'm glad it's getting out their directly instead of being something that's only in the back room that no one's ever heard of outside of a few telcos.
Why is Desktone is building their own datacenter?
When I first heard that Desktone was literally buying their own hardware and renting their own datacenter space, I was thinking, "Why? Why don't you just go to one of the huge providers like Amazon or Terramark and just buy access to their virtual infrastructure? In this day and age, who wants to own hardware?!?"
But of course then I remembered, desktops are not servers! :)
The virtual infrastructure clouds that are out there are designed to run server workloads--databases and web servers and applications--they are not designed to run lots and lots of copies of Microsoft Windows 7. To illustrate this point, one Desktone engineer joked that "Windows is so chatty," and with the infrastructure providers literally charging for each disk IO and byte transferred, "we'd get charged every time a user moved the mouse!"
The other thing to keep in mind is that Desktone spent the last four years creating a software product that runs on bare metal. They need full control of the hypervisor, which is not something that service providers will do. So for that it has to be a private cloud.
Desktone also has the luxury of building for scale. There are a lot of great products in the VDI world (Xiotech, for instance) which are great but that only really make sense in the VDI world when you have a least a few hundred users. But since Desktone is designing for the big time, they can leverage all these fancy new products right from the start.
"Trust us," I've heard from various Desktone employees, "The last thing we wanted to do was to build our own datacenters." But when it comes to desktops, they didn't have a choice--it's either a custom telco datacenter or their own custom datacenter.
One interesting tidbit about this is that right now they're built on ESX. They keep considering XenServer with the tradeoff being ESX's better memory management versus its higher cost. Right now the cost model still keeps them on ESX, but they keep checking to make sure that makes sense.
Scaling up & scaling down
As I wrote previously, Desktone has been designed from the beginning to be a huge, multiuser, multitenant, multi-datacenter package. While today's direct offering is only hosted in a single datacenter (with a single connection broker shared by all customers), Desktone is rapidly expanding to add more locations in the US (for both performance & redundancy) as well as locations in Europe & Asia over the next few years.
Adding more datacenters will allow them to add additional options where customers' desktops & data can be spread across multiple datacenters (which would mean that Desktone desktops were more protected than probably 99% of most businesses' desktops. The whole Desktone platform was designed from Day One to support multiple datacenters and multiple locations. In fact Desktone could eventually put a "datacenter" directly on a customer site, seamlessly extending their cloud to the customers' premise as needed.
VDI? Really? Really!
At one point one of the Desktone folks made a comment that was sort of derogatory towards Terminal Server / RD Session Host. Of course almost by instinct I started to jump on my soapbox in defense of Terminal Server, but then I realized the only real advantage of Terminal Server, all things considered, is the cost. I can spend fewer dollars per user in terms of hardware in my datacenter with Terminal Server over VDI.
Except Desktone is selling VDI. For $30 per month. So given how cheap it is, why wouldn't you want a real Windows 7 desktop with normal apps and admin rights and all the other things you'd expect in a normal desktop? (Of course given Desktone's multitenancy and security requirements, it's also pretty clear why they offer VDI and not Terminal Server hosting.)
My point, though, is that with Desktop, you get VDI for $30 per month. You don't have to suffer through Terminal Server just to save money. I mean yeah I'll defend Terminal Server's honor to the death. But the day that I can do VDI more cheaply, I'm done.
Right now Desktone will support RDP or HP RGS out-of-the-box. But if you've already bought your own Citrix XenDesktop licenses, you can get Desktone to enable HDX for your account too. (Which is great, because it works with all of the standard Citrix Receivers.) Desktone was unclear exactly how this works on the backend (I'm sure due to contractual reasons with Citrix). My sense is that it's more like "HDX Connect" for Desktone, rather than actually using the XenDesktop product.
But this brings up a good point. Citrix is an investor in Desktone, so I'd like to see the two companies work out a deal where HDX can be one of the standard out-of-the-box offerings. Maybe this would be a premium feature, like $5 per month? (Or maybe the two companies could go long and just make it a base feature that's included in the offering. If you want to blow people away with cloud-based VDI DaaS, this is one way you do it!)
This seems like a perfect case for Quest EOP Xtream, although with Citrix being an investor I'm not sure how that would shake out.
2011 will be better for Desktone than 2008, 2009, 0r 2010. But there's still some room for improvement.
Desktone has made a lot of improvements in the past few years, and the new direct approach is exciting. But there's still some room for improvement:
- The Desktone iOS client is $11.99. Yep. In 2011. A VDI company is charging for client software. #fail.
- I'd love to see Desktone partner with a thin client vendor that they could bundle in with the offerings. It would be a nice one-click option to add to your shopping cart when you buy (and of course it would come pre-configured to connect). They could even get into leasing, like $10/month for as long as you have your account, or a free client device if you sign a two-year contract. (Complete with a $150 early termination fee. :) Maybe this is something that an MSP could offer. They could partner with Pano to configure the Easy Button to launch their own helpdesk portal or something.
- I'd also love some kind of referral program where consulting firms or even bloggers with web ads could get something like $1 per month, per referred user, in perpetuity.
- Licensing: Right now if you use Desktone to provide a Windows 7 desktop (they also have Linux options), you have to "bring your own" Windows 7 license since Microsoft doesn't offer Win7 in SPLA. (Don't get me started on that!) So if you don't have your own VDA license, you're looking at an extra $100 per year for your Windows license. (Although Desktone made a smart move and become a reseller so you can buy your VDA license when you're signing up for the service.) But still, I'd love to see a more-integrated pay-by-month offering.
Finally, I heard that Desktone gave at Boston's Virtualization User Group recently. I don't know what they did for the demo, but I know that group only allows two minutes for each sponsor's presentation. If I were Desktone, I would have walked on stage and gone through the whole process of becoming a customer. Visit the site. Create an account. Get the email validation. Click the link. Using my Windows 7 desktop the cloud. Any questions?
I'm very excited for Desktone even though I still think that most users don't need VDI. But if you can use VDI, why not outsource it to someone else who can guarantee your cost? I think Desktone makes a lot of sense for smaller companies. It also makes sense for companies who want to provide a work desktop accessed via whatever existing computing device a user has. (Think of it as BYOC+VDI.) And if you believe with Chetan thinks about how the majority of us will all be using VDI some day, then Desktone makes a lot of sense.
What do you think? $30 a month per desktop? No brainer or fool's errand?