“Will VMware become the next Novell?” and the role of the guest OS
Two questions I keep hearing being repeated are, “will VMware become the next Novell?” along with questions about the role of the guest OS following VMware’s expected purchase of SUSE Linux from Novell. I’ve heard a variety of opinions expressed on this and I felt compelled to offer my own perspective.
Will VMware become the next Novell?
The implication here is that Microsoft will do to VMware (using Hyper-V) what Microsoft did to Novell NetWare with Windows NT/2000 Server.
The premise behind this question is that the hypervisor is what primarily matters and that Microsoft will eventually use Windows Server to “carpet bomb” the market with a high-volume/low-cost strategy and make VMware substantially less relevant.
We need to look at a few things here including the hypervisor today, the hypervisor tomorrow and the role of the hypervisor itself.
Today there are major differences between Hyper-V and ESX 4.1. To pick just one, memory page sharing and memory overcommit has been around since ESX 3.5 and Hyper-V will first get page sharing with Windows 2008 R2 SP1 next year. There’s no equivalent of DRS in SCVMM and vSphere 4.1 just added new features like memory compression, storage I/O control and more. Enough said. While Hyper-V is a capable hypervisor, today there is a significant feature gap.
A picture is worth a thousand words, so here is Gartner’s Magic Quadrant for x86 Server Virtualization Infrastructure for May 2010:
Will Microsoft close this gap in the future? The gap will almost certainly shrink in the long run as all hypervisors will eventually drift toward parity, and that will take some time.
A better question I think is how strong are the API’s and cloud computing ecosystem surrounding the hypervisor?
Consider that VMware has Loadable Kernel Modules (LKMs) for Cisco switches and virtualized application firewalls and agentless antivirus (vShield). Now add the API’s available through the hypervisor for everything from SAN array integration, to the vCloud API and much more. Now look at the suite of products surrounding vSphere –especially vCloud Director. Enterprises will want to choose solutions that enable them to best leverage the benefits of the private cloud and a hypervisor can’t do that all on it’s own.
Microsoft will be a major player in virtualization, but will they be able to dominate the market the way they did against Novell’s NetWare? This is a much different market than the late 90’s and Microsoft would need much more than just a hypervisor with feature parity to do so. In short, I don’t feel that VMware is in much danger of becoming “the next Novell” — and to the contrary I think VMware is positioning themselves to be a leader in cloud computing as well.
The Role of the Guest OS
At VMworld 2010, VMware CEO Paul Maritz expressed a vision of cloud computing where the guest operating system would be less relevant in the long run (i.e. 5 years) in cloud computing environments.
Some felt that VMware’s acquisition of the SUSE Linux OS was a departure from this vision and strategy. In a sense it is, but that future vision of a well-functioning cloud datacenter where the guest OS is less relevant just isn’t a reality yet. In fact, most cloud products are 1.0 versions today.
An acquisition of SUSE Linux may seem to fly contrary to this long term vision, but it could be very valuable over the next 5 years as it gives VMware not only a measure of independence from Microsoft, Oracle, Red Hat, etc., but it also gives VMware the full vertical stack it needs to build cloud ready applications.
To give an example, look at VMware’s recent Zimbra acquisition. VMWare would have a mail platform (Zimbra) with an OS to run it on (SUSE), a virtualiation platform to run the OS on (vSphere,) and vCloud Director to quickly deploy and provision it.
In short, the idea of the demise of the traditional OS is a long term vision that will take years of evolution to develop into reality in our datacenters. VMware owning a major Linux distrubution is a strategic play from which I expect them to significantly benefit over the next 5-7 years until the cloud evolves and matures.
Just please don’t hold me to any 5 or 7-year predictions as my crystal ball is a bit cloudy these days…. 🙂