Today Red Hat announced that it had acquired Qumranet, the company that founded and is the top maintainer of the Kernel-based Virtual Machine (KVM) hypervisor. Qumranet also has an interesting virtual desktop solution called Solid ICE. Solid ICE is a virtual desktop solution that runs on the KVM hypervisor and leverages Simple Protocol for Independent Computing Environments (SPICE) or Remote Desktop Protocol (RDP) for connecting thin clients or remote PCs to desktop VMs.
KVM is a hypervisor on the rise, and it's inclusion in the mainline Linux kernel (2.6.20) certainly has helped its case. Under the hood, KVM has a good architectural foundation. Also, the KVM developers were able to benefit from starting work after many of x86 virtualization's performance and management nuances were brought forth following field adoption of first generation hypervisors.
KVM's architecture and inclusion in the Linux kernel give it potential, but KVM is still a hypervisor that's undergoing many of the same growing pains of hypervisors (e.g. ESX, Xen) that had come before it. I've heard plenty of folks that are eager to dismiss Xen, but Xen has something on its side that KVM lacks - momentum. The open source Xen 3.3 release made a strong statement on Xen's growing maturity and relevance in the market, with performance and availability improvements to compliment the Xen Client Initiative (XCI). XCI's goal is to bring embedded Xen hypervisors to laptops, PCs, and PDAs, and you should expect to hear XCI-related announcements in the coming weeks.
Today's announcement shows me that Red Hat is finally starting to move in the right direction, but work remains. Red Hat feels that they have a chance to be in the top two in the virtualization space, but I think they have a lot of work to do if they want to make that goal a reality. Looking out a couple of years, it's hard for me to see Red Hat as a top four virtualization vendor, let alone top two. To understand the mountain that Red Hat has to clime, consider the vendors that already have a technology and/or market lead over Red Hat:
- Virtual Iron
I think there's a possibility that Virtual Iron may be acquired in the next 12-18 months, so assuming they're out of the picture, Red Hat still has plenty of work to do to supplant VMware, Microsoft, Citrix, Novell, and Sun. Sun, for example, has quietly built out a very compelling virtualization solution (see xVM Ops Center, xVM VirtualBox, and Sun VDI). Novell has also begun to assert itself following its acquisition of PlateSpin and the continued maturation of ZENworks Orchestrator.
Picking up Qumranet is a good step toward becoming relevant in the x86 virtualization space, but Red Hat will have to swallow its pride and take one additional step if they're serious about virtualization - become a Microsoft supported virtualization platform. All of Red Hat's competitors are part of Microsoft's Server Virtualization Validation Program (SVVP), and I see joining the program as a move Red Hat has to make. With a Microsoft support agreement and availability of paravirtualized device drivers that allow Windows OS performance on the KVM hypervisor to be as good as ESX, Hyper-V, or Xen (on Novell, Citrix, Virtual Iron, or Sun), Red Hat will have a chance. More than 85% of x86 VMs run a Windows OS, so to be a mainstream x86 virtualization player, Red Hat has to offer a Microsoft-supported solution. Otherwise, Red Hat's virtualization offering may mature to the point that it is a good Linux virtualization platform, but it will never have the breadth to seriously challenge its chief competitors and be a top 5 virtualization vendor.
Red Hat - like it or not, a strong solution for Microsoft operating systems is your ticket to the virtualization party. Until you have that ticket, you will remain an outsider peeking in the virtualization party window.
Posted by: Chris Wolf