The enterprise adoption of virtual desktop infrastructure technology has fallen short of market expectations, leaving vendors to wonder what the barrier to growth is. After all, VDI is proven and is dropping in price, so what piece of the adoption equation is missing?
In a word: scalability. But it's not a question of scalability in the traditional sense, where the term refers to making systems larger. It's more of a question of scaling out to remote locations, branch offices and mobile workers. That type of horizontal growth introduces many more variables into VDI viability, factors that include performance and the end-user experience.
Multiple components affect virtual desktop infrastructure (VDI) performance -- server design, hypervisor capabilities, virtual machine density, display protocols and bandwidth. Administrators can use a localized deployment to gain some control over each of those components to scale VDI. However, bandwidth becomes a variable element once VDI moves out over the WAN -- especially if the Internet is the primary transport mechanism.
Unpredictable WAN performance, high latency and limited throughout are major considerations when looking to push virtual desktops over the WAN. And that is exactly where display protocols can demonstrate their efficiencies and merits. Let's take a look at the major display protocols and how each one handles VDI sessions over a WAN connection.
Microsoft's Remote Desktop Protocol (RDP) has been around since the days of Windows NT 4.0 and is now in Version 7. RDP 7 includes many improvements specifically engineered for Windows 7 environments. For example, RDP 7 includes native support for Aero Glass, multimonitor support and runs over TCP/IP, making it suitable for routing via the Internet.
RDP is a multichannel-capable protocol, allowing for separate virtual channels for carrying presentation data, serial device communication, licensing information, and highly encrypted data such as keyboard and mouse activity.
The sending and receiving of data through the RDP stack is essentially the same as the seven-layer Open Systems Interconnection model standards for common local area networking today. Data from an application or service to be transmitted is passed down through the protocol stacks, sectioned, directed to a channel, encrypted, wrapped, framed, packaged onto the network protocol and finally addressed and sent over the wire to the client.
The returned data works the same way, only in reverse, with the packet being stripped of its address, then unwrapped, decrypted and so on until the data is presented to the application for use.
Key portions of the protocol stack modifications occur between the fourth and seventh layers, where the data is encrypted, wrapped and framed, directed to a channel and prioritized. However, RDP is not optimized for WAN support or to work with high-latency connections. RDP exhibits performance problems over connections with high latency or contention, especially when it comes to transferring video (screen/display) updates. That has led to third parties developing acceleration technologies that use data de-duplication, caching and more efficient encryption schemes to move RDP traffic over the WAN. That acceleration technology is available either as part of a connection broker or a WAN optimization solution. Either way, for RDP to offer acceptable performance over a high-latency WAN, a third-party product is usually needed.
Independent Computing Architecture (ICA) is a proprietary protocol for an application server system designed by Citrix Systems. The protocol lays down a specification for passing data between servers and clients, but it is not bound to any one platform.
ICA can tune the TCP window size to improve performance. In addition, the protocol incorporates a number of compression techniques, including bitmap image compression, screen-refresh compression and general data compression, all of which helps to improve the end-user experience over a WAN connection.
However, just as with RDP, contention and latency still have a major effect on the quality and usability of video over the WAN. To improve the end-user VDI experience, Citrix has introduced its HDX technology. HDX isn't a single product, but a branding of several products, each with the goal of improving ICA's abilities to deliver data to a thin client. Citrix has created several sub-brands under HDX, such as HDX MediaStream, HDX RealTime and HDX 3D. Each of these incorporates a variety of technologies. For example, HDX Plug-n-Play supports client USB devices, multiple monitors, client printers, client drive mapping, local port mapping, smart cards and scanners.
The ICA/HDX offering from Citrix offers some enhanced support for high-latency connections, but as with many other display protocols, ICA/HDX will benefit from WAN optimization and acceleration technologies.
PC-over-IP (PCoIP) was originally designed by Teradici for remote computing/zero-client solutions. Teradici's first product to use PCoIP had little to do with virtualization; it consisted of a "display card" installed into a PC in the data center. The proprietary card would encapsulate all of the PC's functions into TCP/IP traffic and then transmit them to a remote thin-client device (or PC) using PCoIP. This provided remote computing to the endpoint, using the LAN as the transport mechanism. Today, VMware uses licenses the PCoIP protocol for its VMView technology.
The PCoIP display protocol functions differently from other display protocols and incorporates technologies that improve performance over WAN connections. For example, PCoIP is designed to recognize the content type and apply the most efficient compression algorithm for that content. Teradici claims that the technology, as well as other transport performance enhancements, can help reduce display latency by more than 50% for common tasks when compared with legacy display protocols such as RDP.
PCoIP incorporates several other technologies that work together to improve performance over high-latency, low-bandwidth connections. For example, host-rendered pixel encoding places the burden on the host device to render images, making screen updates independent from latency and bandwidth issues. Host rendering also allows PCoIP to use multiple imaging coder/decoders (codecs) concurrently, selecting the most efficient codec for each type of image. PCoIP also uses a "progressive build" methodology for screen updates to improve performance. Progressive build works by initially populating the screen with a highly compressed "lossy" image and then progressively building the image as bandwidth allows. Those capabilities and others help PCoIP perform over WAN links for VDI implementations. WAN optimization and acceleration offers some help for PCoIP, although the level of performance improvement may be less than with other display protocols, simply because PCoIP already has several features for maximizing WAN performance.
Beyond display protocols
Effectively deploying VDI over WAN connections is not really a question of which display protocol to use, but more of which companion technologies exist to maximize VDI/WAN performance. Perhaps the best approach is to think of VDI as any other technology that can traverse the WAN, where latency and bandwidth have a major impact. That said, any technology that can boost WAN performance can benefit those implementing VDI.
ABOUT THE AUTHOR
Frank Ohlhorst is an IT journalist who has also served as a network administrator and applications programmer before forming his own computer consulting firm.