InfoWorld Test Center Looks at Four Desktop Virtualization Solutions

InfoWorld’s Test Center and Randall C. Kennedy recently published their findings after trying out four competing desktop virtualization solutions from Microsoft, Parallels, VMware and new comer InnoTek.

Based on his testing, Randall scored the products like so:

• InnoTek VirtualBox 1.3 – Received a Fair rating and a score of 6.9

• Microsoft Virtual PC 2007 – Received a Good rating with a score of 7.4

• Parallels Workstation for Windows 2.2 – Received a Good rating with a score of 7.8

• VMware Workstation 6.0 Beta 3 – Received a Very Good rating and a score of 8.3

Randall concludes the article with the following thoughts:

As with most mature market segments, the major players in the desktop virtualization space have carved out sustainable niches for themselves. VMware Workstation continues to dominate the developer landscape, with Version 6.0 further solidifying its position at the top of the virtualization heap. Microsoft’s Virtual PC 2007 has been relegated to a “VMware player” type of existence, while Parallels Workstation for Windows 2.2, though a speed demon, struggles to escape the shadow of its favored (and better equipped) sibling, Parallels Desktop for the Macintosh. And despite a poor benchmark showing, VirtualBox still delivers a modular, developer-friendly architecture that effectively complements its open source licensing.

Longer term, these vendors have one of two options: Dig in and try to carve out the biggest piece of a very limited pie (developers and help desk professionals); or evolve their products into something more all-encompassing, as VMware has attempted to do with ESX, VDI, ACE, and other acronyms du jour. With Microsoft already focused on the datacenter and with InnoTek well on its way to delivering an ESX-type solution, that leaves Parallels as the odd man out.

And although VMware’s product achieved the highest rating and score, VMware’s John Troyer still wanted to set the record straight on two points of the article. First, John questions the benchmarking of a beta build of Workstation. And I suppose he has a point here. Although VMware came out on top, any performance tests made on a beta build of VMware’s product would be hurt by the fact that the beta build is in debug mode. However, it probably wouldn’t have been right to test with VMware Workstation 5.5. And secondly, John questions Parallels’ use of the word hypervisor to describe their workstation product. John offers his own definition of hypervisor and then tries to explain what he believes is a better explanation of Parallels’ approach.

You can read John Troyer’s response on the VMTN Blog site, here.

And you can read the entire InfoWorld Test Center article which contains all of Randall’s findings, thoughts, charts and “Bottom Line” information, here.