Revving Up Desktop Virtualization

Arthur Cole

Amid all the hype over desktop virtualization, there is one simple fact that tends to be overlooked: Virtualizing a desktop is a lot more complicated than virtualizing a server.

In the server farm, a single device gets virtualized and, for the most part, the rest of the enterprise simply sees two or more servers in its place. Optimizing that environment requires you to branch out into the server backplane and onto the network and even into storage, but nowadays a basic virtual system can be up and running with little fuss.

On the desktop, however, we quickly run into all kinds of complications, brought on largely by the fact that the desktop image is not only undergoing virtualization but is being relocated back to the central processing architecture where it is parceled out to stripped-down clients throughout the organization. So just to get your foot in the door, you're looking at reworking everything from the desktop to centralized server and storage architecture, plus multiple points on the LAN and even out over the WAN.

At the moment, the chief problem to overcome is latency. After generations of instant response between mouse/keyboard to the local processor and back, users just can't tolerate a slow desktop. That's why we're seeing a raft of solutions aimed at accelerating virtual desktop architectures. FastSoft, for one, says it can deliver a five-fold increase in thin-client applications like Citrix Remote Desktop and Microsoft RDP over wide area infrastructure, without having to install client-side software. That means it can cut the download time of typical PDF documents from five seconds to one second or less.

Microsoft is keenly aware of the latency problem as well, and is not about to let an upstart like FastSoft encroach upon a market that it covets. Redmond has long known that its RDP protocol has lagged behind the technology of chief competitor VMware, so it has issued a new accelerator to speed things up. Based on technology acquired from Calista, the new RemoteFX module is limited to users of Windows 7 SP1, but is compatible with both RDP and Citrix' HDX format.

And newcomers like Red Hat are hoping to hit the ground running with revved-up desktop capabilities right out of the gate. The company has integrated the once-separate desktop edition into the Red Hat Enterprise Virtualization platform, adding the Simple Protocol for Independent Computing Environments (SPICE) format it gained from Qumranet last year. The system provides a connection broker and a Web portal that allows remote users to access their desktops from virtually anywhere.

The sudden push to speed up in-house VDI platforms make sense when you consider the new breed of hosted solutions that are coming online. Rackspace recently launched a new desktop suite powered by Virtual Bridges' Verde platform that offers up to 1,000 seats without any of the capital costs of a typical deployment.


Despite the sudden rush of acceleration technology, virtual desktops won't be able to match the traditional desktop in responsiveness any time soon. But lightning-fast desktops may be a luxury for many knowledge workers before too long. Top management is already well aware of the economics behind the virtual desktop, and they make a very persuasive argument.

Add Comment      Leave a comment on this blog post
Mar 31, 2010 4:08 AM Jim Curtin Jim Curtin  says:

While the Virtual Bridges offering with Rackspace addresses the capex issues around VDI, the majority of companies are still looking at on-premise deployments initially. Latency is a huge issue. Virtual Bridges addresses the latency in several ways - first, the VERDE protocol. Where that is not sufficient for reasons of WAN latency, we have Cloud Branch which distributes the processing down to the LAN at the branch level without adding any management responsibility to the branch. This gives a native PC experience to users in even the most remote branch. Another key differentiator is out client-side hypervisor for disconnected use. Regardless of the network latencies, you always have a fully-managed PC at your fingertips. This combination of VDI plus disconnected use/CSH plus Cloud Branch gives users multiple ways of solving the network latency issues that are keeping other VDI deployments stuck in the HQ...

Mar 31, 2010 9:58 AM Daniel Feller Daniel Feller  says:

You are correct in that desktop virtualization, or VDI as some people refer to it as, is more complicated than server virtualization.  It goes well beyond the area you focus on with regards to latency. 

In addition to latency, which fits into a larger issue (user topology), you also have to content with standards, migration, applications, users, etc.  I've been discussing this in a blog series called Desktop Virtualization... The Right way (

However, like any project, if you do it right, if you plan properly, if you don't expect the impossible, desktop virtualization has some real benefit.

Mar 31, 2010 11:44 AM Martin Ingram Martin Ingram  says:

Much of the latency you mention is down to display protocol used to tranfer the user interface from the data center to the user. There is a huge amount of progress being made in display protocols at the moment and my expectation is that this will become a non issue for the majority of users over time.

Your comment on the complexity of desktop virtualization in comparisson to server virtualization is well made but this actually refelcts the complexity in herent in most desktop dxeployments today. Desktop virtualization ultimately allows us to manage desktops in a far more efficient manner by keeping seperate the OS, applications and user environment and managing each in the most appropriate way.

Martin Ingram (AppSense)


Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.