Virtualization: The coming display protocol wars
The future of desktop virtualization may hinge on display protocols
- By Tom Valovic
- Jun 26, 2009
Back at its industry conference in September 2008, VMware Inc. had its work cut out for it. With new CEO Paul Maritz having just replaced Diane Greene, the company had to come out with guns blazing in front of an impressive crowd of 14,000 attendees. For the most part, the company managed to pull it off with a slew of announcements at the show -- and in the weeks that followed, which are usually an informational dead zone.
Nearly lost among those announcements was a licensing and co-development agreement with Teradici Corp., a small Canadian company of which most attendees had never heard. But as the virtual desktop infrastructure (VDI) market begins to gather steam and VMware devotes more and more resources to it, the Teradici deal will loom larger, becoming an important strategic puzzle piece in the emerging VDI protocol wars.
All About the Network
As a desktop solution, VDI is network-dependent. In one type of common configuration, a virtual server in the data center communicates with a rich or thin client on the end user's desktop using a display protocol across the LAN. Because a network sits between the end user and the compute resources, the protocol's job in part is to optimize this process as bitmap changes are sent to the end-user client. The more efficient the protocol, the snappier the end-user experience.
For this reason, display protocols are critical to the overall success of VDI -- a fact that hasn't been lost on the top VDI vendors. Companies such as VMware and Citrix Systems Inc. are keen to make sure that early VDI implementations don't frustrate end users, who are used to the speedy performance provided by today's laptops and PCs.
Today's mainstream display protocols are well known and largely proprietary. They include the Citrix Independent Computing Architecture (ICA); a WAN-optimized protocol from Sun Microsystems Inc. called Appliance Link Protocol (ALP); Microsoft's Remote Desktop Protocol (RDP); Remote Graphics Software (RGS) from Hewlett-Packard Co.; and Smart Protocol for Internet Cellular Exchange (SPICE) from Red Hat Inc., which acquired the protocol when it bought Qumranet.
Among the more widely used protocols, Citrix ICA is considered to be the best performing, in no small measure because Citrix has had 15 years to perfect it. It's long been a key element in the company's presentation-virtualization solutions, such as XenApp, formerly known as Presentation Server.
As VDI becomes more mainstream, VDI suppliers will be honing these protocols to improve their existing product offerings because the bandwidth demands on IT networks are increasing rapidly.
For example, the use of multimedia has been growing steadily over the years and shows no signs of abating. This includes apps such as video streaming, videoconferencing, VoIP and unified communications. Some of these applications are isochronous, meaning they function in real time and are intolerant of delay. The challenge for both thin clients and VDI will be to catch up with these kinds of applications by offering the same level of performance available with a PC.
VMware's VDI solution is called VMware View. Traditionally, VMware VDI products have used Microsoft's RDP. But RDP can't match the performance of Citrix ICA, which is why HP recently announced it's working to address some of the issues that caused RDP to lag behind ICA in performance.
HP plans to build extensions to RDP to improve performance for graphics and real-time applications such as videoconferencing and VoIP. Some improvements have already been incarnated into a new software suite, called HP Virtual Client Essentials.
Microsoft is also working to improve RDP. In January 2008, it acquired Calista Technologies, a start-up focused on presentation virtualization. Calista's technology is based on a compression algorithm that optimizes the RDP protocol and requires less network bandwidth.
VMware is taking a different tack to improve the display protocol aspect of its VDI offerings, starting with the Teradici deal. Teradici has developed a unique approach to display protocols and is attempting to push the envelope for VDI with a technology called PC-over-IP (PCoIP), which uses unique graphics algorithms, silicon processing and host workstation/server add-in cards.
In essence, Teradici is VMware's attack dog in the protocol wars. In part, this is because Teradici says that PCoIP has features no other VDI protocol offers, such as an embedded optimization algorithm for dynamic adjustments to network performance.
Stuart Robinson, director of business development for Teradici, claims that ICA simply can't do the same. "If there's an instantaneous change in the network, we can reflect [that] change in the quality or frame rate. That allows us to be very responsive from a user perspective, even if the network is fluctuating," he says.
Citrix CTO Simon Crosby disputes the claim that ICA can't make dynamic adjustments. Writing in his blog -- in response to a blog that appeared on the Virtualization Review magazine Web site -- Crosby said: "Citrix ICA has had an ability to dynamically adapt its coding algorithm not only in response to availability of network bandwidth, but also in response to available encode/decode capabilities at the server and client side, respectively, for over two years. This allows ICA to deal with complex rendering problems with grace and to deliver high fidelity across a highly disparate set of server/network/client combinations, dynamically adapting as system conditions change."
VMware's main VDI competition comes from Citrix rather than Microsoft, so there are important issues at stake because Citrix has an array of products it sells to address LAN/WAN optimization.
Software, Not Hardware
Take Teradici's PCoIP and convert it to software, and you have the essence of what the VMware partnership is all about. Under the agreement announced at VMworld, VMware will effectively license a software implementation of PCoIP. The protocol will be included in a future release of VMware View, allegedly enabling such applications as streaming movies, high-resolution graphics and high-definition audio.
According to VMware's Senior Director of Enterprise Platforms Jerry Chen, the companies will collaborate to convert Teradici's chip-based hardware solution into a similar software offering. This is something at which VMware, as a pioneer of hardware virtualization, excels.
This software-only version will eventually be embedded in VMware View as well as thin clients, in keeping with Teradici's server/client approach. But Citrix's Crosby says that, based on conversations he's had with OEMs, "That approach is proving to be a difficult sell to customers, since it ties the server forever to a particular delivery protocol from a particular vendor in a narrow proprietary architecture."
Analyst Rachel Chalmers, with The 451 Group, questioned in a report whether the software version of PCoIP could get the job done. "A software PCoIP won't offer the same performance as its hardware equivalent, but it would give VMware a workaround for Microsoft's ubiquitous-but-pedestrian RDP. That could become an important differentiator for VMware," she wrote.
Citrix, in the meantime, is not sitting still. The company is spending research and development effort on Project Apollo, which centers on 3-D graphics as well as streaming media and real-time capabilities. The project will have applications to both XenDesktop -- Citrix's VDI solution -- and XenApp via an approach called SmartRendering, which combines server-side and client-side rendering to improve desktop performance.
VMware says it will also work with Teradici on further network-optimization capabilities for high-latency WANs and other network environments. Chen says that with these refinements, it's possible that RDP could be replaced. Whether VMware will succeed in this effort remains to be seen, and some observers feel that the process could take years to develop.
The VDI Battlefield
The race is on. But the overall success of VDI will ultimately depend on the judgment of millions of users, who are used to multi-core horsepower on their machines with no intervening networks to add latency. If VDI ultimately represents a step backward, then it will have a difficult time taking hold in the marketplace. The outcome of the protocol wars may be a large factor in how it all shakes out.