Red Hat's desktop virtualization software and its acclaimed remote desktop protocol reproduce as much of the native desktop experience as possible, from multimedia playback to locally connected hardware. How far the open source company can carry its innovations is another story, though -- since its competitors deliver similar solutions and have a much larger customer base.
Red Hat's virtual desktop vision, constraints
The idea behind
The advantages of RHEV for desktops are the same as with other desktop virtualization products. You can provision or de-provision instances of a desktop system on demand -- you're not confined to any one particular location for a user's system and the virtualized systems are all managed centrally. End user machines are also completely isolated on the back end server and Red Hat uses page-sharing and other resource optimization algorithms to get that many more virtual machines running side by side on the same hardware.
But the drawbacks of server-hosted desktop virtualization are hard to ignore. In general, the end-user experience is quite limited (video and sound are usually terrible), the users are limited by the local hardware used to connect to the remote server and the whole experience is further constrained by the available bandwidth and the latency of the network. In plain English, there's only so much that can be shoved across the network at once.
Red Hat's secret sauce: SPICE
Red Hat's answer to virtual desktop performance challenges is SPICE -- an open source network protocol originally developed by Qumranet and placed under Red Hat's stewardship when it acquired the company in 2008. SPICE is one of Red Hat's "emerging technologies" in the virtualization space along with KVM (the virtualization system used within Red Hat Linux) and Red Hat takes the development of this technology seriously.
SPICE works by creating several generic interfaces or "channels," each for a different aspect of the user experience and each implemented in a highly abstracted fashion so that it works across many platforms. Display actions, user-interface device inputs, cursor movements, audio playback and audio recording each have their own dedicated channels. The guest operating system needs to have special drivers installed to relay and accept actions through SPICE, but that's not terribly different from the "guest extensions" installed in most every virtualized instance of an operating system.
The SPICE client has its own bit of magic, as well. USB hardware on the client side can be remotely attached to the virtual machine, so devices like printers or scanners can work as expected. Multiple displays are also supported. And because the protocol has dedicated extensions for multimedia, sound and video playback can also work on the client that much more effectively.
The client software also has extensions to allow graphics processing to be offloaded to the client if the hardware there can support it (e.g., video decompression). The protocol's heavily open sourced, which makes it that much easier for new client modules or guest extensions to be developed over time.
Why SPICE might not be so savory
If SPICE is so great, what keeps it from being widely adopted? The first and biggest obstacle is the inertia of existing remote desktop technology deployments -- Citrix, VMware, Ericom and Microsoft already have big chunks of the market sewn up.
Citrix is the market leader (with ICA/HDX), but VMware View's most recent incarnation of PCoIP (version 5) is competitive. Microsoft recently upped its game with the introduction of its RDP add-on, RemoteFX. It publishes high-performance video and audio and USB redirection from Windows 7 guests hosted under Windows Server 2008 R2. Using this, it's not hard to see how a "Microsoft-to-Microsoft" setup would have a native performance advantage over more heterogeneous solutions.
That said, not all businesses can afford those technologies and not all companies want to be locked into those technologies. SPICE supports a more mix-and-match approach, where the client environment and the guest OS can be, in theory, anything.
But, SPICE isn't a polished solution, yet. Many of SPICE's underlying features are only partly implemented (e.g., smart cards or USB support) or have to be added manually by third parties --as was the case with Virtual Bridges VERDE 5, which implements SPICE in a custom-written stack.
Consequently, people who need a fully developed solution today are more likely to commit to one of the existing, proven solutions rather than one that's still evolving and largely untested in the real world.
One could argue that the state of SPICE is a byproduct of it being open source -- albeit derived from a proprietary endeavor. So, while Red Hat's desktop virtualization technology is competitive with what's already on the market, the biggest challenges Red Hat faces is with compelling potential customers to switch away from their existing, proven solutions. That's at least as much about marketing strategy as it is the technical merits of RHEV.
ABOUT THE AUTHOR:
Serdar Yegulalp wrote for Windows Magazine from 1994 through 2001, covering a wide range of technology topics. He now plies his expertise in Windows NT, Windows 2000 and Windows XP as publisher of The Windows 2000 Power Users Newsletter and writes technology columns for TechTarget.
This was first published in September 2011