Is it time to fork Xen?

I read this Charlie Babcock (InformationWeek) interview with Peter Levine (CEO, XenSource) and Wes Wasson (corporate VP of worldwide marketing, Citrix), and I'm left with the feeling that XenSource really doesn't see Xen's future in open source, but rather in Windows.

Which makes me think it may be time to fork the project and move on. But then, the company already did that for us, didn't it?

As Michael Tiemann writes, one of the wonderful attributes of true open source is that it divorces you from the need to trust your vendor. If the vendor zigs (down an "evil" path), you can zag. Read these comments and tell me what you think should be done with Xen/XenSource:

INFORMATIONWEEK: Is the work being done with Microsoft separate from the open source work?

Levine: The reason for an engineering office in Redmond is our strategic relationship with Microsoft. It started over a year ago, before any talks with Citrix. It was crystal clear to me that the bulk of the virtualization market was going to be among Windows customers. Red Hat and Novell had already embedded Xen in Linux, and Linux was going to be a much smaller market.

Wasson: We want to preserve a great relationship with Microsoft [Matt note: Of course Citrix does, as most of its business is Windows-related], one that's pretty unique. What XenSource is doing with Microsoft in virtualization mirrors what Citrix did in Windows Terminal Services. We helped extend the Microsoft product. We got the right stuff [Citrix Presentation Server and proprietary ICA protocol] into their environment. For every dollar of Presentation Server we sell, Microsoft gets 75 cents.

We gave code to Microsoft that became part of the core of Windows Terminal Services and got source code rights to Windows Server. We work with Microsoft on things that are years in advance. Microsoft is tremendously willing to do that when you're not pretending that things that should go into the operating system [such as the virtualization hypervisor] are not available to them and will be supplied somewhere else. VMware is thinking of itself as an operating system vendor and Microsoft competitor. They want to compete with the Windows operating system.

In other words, XenSource's future with Citrix is Windows and helping Microsoft. Sounds like a fork-worthy situation to me.

Fortunately, XenSource has basically already done this, separating the Xen project (and governance thereof) from XenSource, the company (and the proprietary work it's doing). Thank you. That makes it easy.

What's sad is how little value XenSource seems to put on the Xen project, despite the fact that no one would care $.02 (much less $500 million) for XenSource but for the Xen project. Peter Levine feels the same, but for a different reason:

"While the engine [Xen project] is open source, the car is the value-add that customers need," said Levine, referring to XenSource's value-add virtualization services and management capabilities, which are not open source.

This would be more credible if customers were paying XenSource much for its "car." They weren't/aren't. The company has a big OEM deal coming that is expected to account for much of its 2008 revenues, but its current customers pay $2,000/each (roughly) for the car. It sounds to me like XenSource had a hard time building more than a Yugo around the engine.

But that's its fault, not Xen's.

Microsoft may be the future of XenSource (rumors are swirling that Citrix is an acquisition target, which makes sense given how much of its business is tied up in Redmond), but Microsoft is not the future of Xen. We can thank XenSource for recognizing its allegiance early and setting Xen free. Of course, the GPL already did that, which is probably why Microsoft has no great love for the project anymore.

Tant pis. Xen is a great project with a great future. I would have liked to have seen XenSource help steer that future, but it will be busy working on Microsoft's. Life goes on.