Michael Dolan Dot Com

Friday, May 9th, 2008

An honest look at Xen and a future with KVM

Anthony Liguori has an excellent blog post up taking a very honest/hard look at Xen and what likely lies ahead (KVM).

I think we can finally admit that we, the Linux community, made a very big mistake with Xen. Xen should have never been included in a Linux distribution. There, I’ve said it. We’ve all been thinking it, have whispered it in closed rooms, and have done our bests to avoid it.

He makes a very valid point that the Xen kernel will never be upstream and therefore it will always be an extra layer. The difference between Xen or other packages around the kernel is that Xen is at the core its own kernel. This is what makes KVM so attractive - you don’t need that extra layer (which creates complexity no matter what).

Quite simply, Xen is not, and will never be, a part of Linux. Therefore, including it in a Linux distribution has only led to massive user confusion about the relationship between Linux and Xen.

Now one thing Anthony didn’t mention is that the ability to swap Xen as the virtualization technology with another, future, and better technology was something the kernel developers perceived long ago. Because of that foresight, most people are starting to use libvirt which abstracts the interface layer into managing virtualization hypervisors like Xen and KVM. What does this mean for enterprise and ordinary users alike? The migration from one technology like Xen to another like KVM can be done near seamlessly because the libvirt-based management tools can plug into either hypervisor, checkpoint start/stop, and bring the images up on another hypervisor.