When Red Hat announced its decision to switch virtualization technology, moving from Xen to KVM, in June 2008, it generated a lot of buzz.
It was a dangerous move, considering that the platform was pretty new, that its creator and maintainer was a young startup, Qumranet, and that no ISV was actually supporting its applications inside it.
On the other side KVM was integrated in the Linux kernel after just six months of development, and Red Hat eventually acquired Qumranet to get the knowledge, the people and the influence to return the most on its risky investment.
Nobody followed Red Hat: Citrix, Virtual Iron, Oracle, Sun and of course its primary competitor Novell continued to work on Xen.
Fast forward to late 2009: Red Hat is finally ready to unveil its commercial implementation of KVM, introducing Enterprise Linux (RHEL) 5.4, Enterprise Virtualization Hypervisor (REVH) and Virtualization Manager for Servers (REVMS).
Red Hat continues to be the only virtualization player to offer a commercial implementation of KVM, but but things may change soon.
Novell is in fact researching a new hypervisor built on KVM called AlacrityVM:
AlacrityVM is a performance focused hypervisor based on the Linux KVM project. Virtualized environments often impose significant performance penalties against a given workload when compared to native “bare-metal” equivalents. This project is motivated by the belief that it doesn’t necessarily have to be this way, nor do we need exotic hardware to achieve it. AlacrityVM demonstrates that most of the existing performance bottlenecks in today’s system are simply the result of suboptimal software stacks. By systematically identifying and fixing the weak links in the guest/hypervisor equation, near native performance from a virtualized environment is realistically achievable.
We also aim to add new features, such as the ability to express real-time constraints, network qos, virtual filesystems, etc.
AlacrityVM is in a very early stage (the first public build appeared in March 2009) but it seems that Novell already submitted it to the Linux maintainers for inclusion in upcoming kernel 2.6.33.
Linus Torvalds rejected it saying that there was not enough interest around the project.
Part of the kernel or not, it demonstrates that Novell is considering and investing on Xen alternatives. And this is particularly interesting considering that the company announced its plan to release a lightweight version of their virtualization platform (read: with a stripped down version of SUSE Enterprise Linux) in March 2008 and never delivered so far.