Today we announced another key addition to our portfolio of HDX technologies within XenDesktop, enabling high definition virtual desktops – HDX 3D.
Before we get into too much detail, in my experience it’s always sensible and logical to consider the broader context to gain perspective and not get lost in the marketing. For me this is about extending our leadership position in delivering user experience that we have built with HDX technologies over the last 20 years.
HDX offers several other capabilities such as multimedia, USB, voice collaboration, etc that users for virtual desktops can use wherever they are. Michael Harris produced a really nice short video that explains the HDX areas of competence. What’s key to understand is that to deliver the best experience for many rich content types, on diverse devices that may be accessed from a myriad of locations, one must leverage the existing environment in an optimal manner. HDX with adaptive orchestration leverages the client, network and server to optimize the experience of users based on the content and infrastructure that is available.
This inherent flexibility in the architecture of HDX technology is key to delivering a more efficient and predictable user experience. For example if you send multimedia content directly to the client in it’s original format for optimal playback when the bandwidth is good, and render on the server when bandwidth is poor you leverage your bandwidth efficiently allowing other applications to also consume bandwidth. Similarly if your technology can identify repetitive patterns within multiple bitmaps on a screen you can avoid retransmitting bits saving network bandwidth. Additionally if your technology can dynamically adapt to changing network conditions, you remain efficient as parameters that you can’t control change. These unpredictable conditions become even more important as you traverse the internet to connect to clouds. So having an arsenal of technologies to handle so many variables that have been matured over the years is critical to delivering a user experience that will allow IT to feel confident in the service levels they are delivering for Desktop Virtualization.
Extending HDX to graphical intensive applications over any kind of network- HDX 3D
HDX 3D is a new technology that addresses the needs of professional graphics users in industries such as Engineering, Manufacturing, Oil & Gas, Aerospace and Automotive. Professional graphics users requiring 3D graphics apps such as AutoCAD or Catia can now use virtual desktops to do all their work.
When we collaborated with our customers on the design of this technology, they reminded us that with the forces of globalization in play, more and more designers and engineers would be working offshore all over the world. This made it abundantly clear to us that in order to meet customer needs WAN performance was going to be a critical component. If WAN optimization was not a top priority from the outset, many of the cost benefits would be negated through poor performance and increased WAN circuit costs. So the mission for our engineers was very clear.
Our test engineers reported back that they could deliver a great user experience on WAN connections with latency of up to 150ms. Even 3D graphics over 1 Mbps became possible! I was pleased, but wanted to understand how this compared to hardware based solutions that have generated a lot of excitement in the industry. Our engineers decided to contrast performance against the PCoIP protocol leveraging the proprietary Teradici hardware solution. Note that VMware has publically stated that they plan to offer a software only version of the PCoIP protocol. Logically speaking, the same technology without the benefit of hardware assist, one would expect this version to be of inferior performance. To be fair, we have not tested the software only version of PCoIP yet.
The methodology on the LAN was to auto-spin in Catia and measure bandwidth usage starting at full image quality for each product and incrementing down to the lowest quality. For the WAN, a WAN emulator was used to introduce latency and repeat the test to auto-spin Catia and observe which appeared to spin more smoothly. Manual rotation was also performed to see what it felt like from a user perspective.
In summary we found that PCoIP with the benefit of proprietary hardware consumed 10X more bandwidth than HDX 3D and HDX 3D produced a smoother user experience.
Based on our tests, PCoIP didn’t handle latency and WAN conditions well. So, if you are a designer and need to design a 3D model, from our tests we don’t see how you would be able to have a predictable work experience that is efficient and cost effective on a real world network with a distributed workforce. Once again, this was the proprietary Teradici hardware assisted version of PCoIP. This does not bode well for the software version, especially if compromises have been made in quality and predictability to mask bandwidth requirements.
So what should one make of all this?
Firstly, as a technologist, I think it’s interesting to see VMware building (or OEM/partnering for) their own protocol; because as a former customer of VMware, I have never understood their position on recommending multiple protocols for different scenarios. They have mentioned extending Microsoft RDP, using Suns ALP for WAN, adopting the Net2Display standard and partnerships with hardware vendors such as Teradici etc. How is all the complexity of so many options and expecting customers to figure out how to integrate them as one predictable experience going to work? Now I wonder if VMware is still going to have multiple technologies to try and build a HDX like portfolio and expect customers to figure out how to interface it all? PCoIP is only a piece of a puzzle – just like a protocol is only a small part of the equation.
However as Citrix has learned over the last 20 years as one of it’s core competencies working with 200,000 plus customers, it’s so much more than just a protocol. Some of those key lessons are.
• HDX is an integrated set of technologies that are intended to offer the best possible user experience under any network circumstances.
• HDX does not rely on just one approach to graphics and multimedia remoting, because one size does not fit all.
• HDX has unique technologies that can leverage the client resources to offer a local like user experience at a fraction of the bandwidth and server cost.
• When HDX determines that server side rendering is the most appropriate method to deliver the content, it uses a number of technologies to optimize the bandwidth and server CPU usage.
• HDX also includes adaptive orchestration which is a system that makes smart decisions on what techniques to apply under different conditions. Adaptive orchestration sets HDX apart from the competition because it can offer the best possible user experience for the user.
Additionally, I think one of the most important points that is often overlooked is that HDX technologies are Hypervisor and hardware agnostic. Citrix does not lock you into a proprietary hardware solution or Hypervisor. We understand that customers want Hypervisor and hardware choice combined with mature and proven user experience delivery technologies that are being extended to further the reach of Desktop Virtualization. I believe that pragmatic IT decision makers and engineers understand those are critical elements to protect their existing and future investments.
Finally, we would have demonstrated HDX 3D at VMWorld, if we could – if we were allowed the space to do so…… However check out http://hdx.citrix.com