The 'Virtualization Wars' Heat Up
Remember the Unix wars of 20 years ago? When it began back in the mid-1980s, it seemed all too clear which operating systems would lead the the computing world into the 21st Century. After all, the University of California, Berkeley was behind BSD, and AT&T backed System V. Virtually Speaking: Remember the Unix wars of 20 years ago? Brace yourself for the next big one: the virtualization wars.
Today, BSD still manages the occasional toot, but when was the last time you heard from someone running System V?
In fact, this decade, as x86 virtualization has picked up steam, it's become less about the OS, and more about the virtualization environment: the hypervisor and beyond. And, as the market continues to come to a boil, it's become increasingly clear, especially in recent months, that the technology may be turning into this decade's equivalent battleground minus the zealots, who may or may not arrive to fan the flames.
There are, of course, key differences (e.g., environments can coexist and there is no inherent intent of collaboration), but the striking similarity is that both represent a paradigm shift in the architecture of the underlying computing infrastructure.
Schorschi Decker, editor-at-large of ToutVirtual's "A proper virtual World" blog, recently offered some insight and predictions about the future of the virtual landscape in a blog post titled, "Retrospective, What is New is Old?" The entire entry is worth a read, but here's what he had to say about today's leading environments:
Xen will not survive, not because it is not good, it is, but because market share trumps all. Citrix has not owned the low-cost virtualization market segment, and now with other competition maturing they are under severe threat. Xen also is not free and compared to KVM and Hyper-V is not that cheap, at least not as cheap as one would wish.
Hyper-V will struggle for another two years, because it is still weak. Being free only gets you so far. Microsoft will make Hyper-V successful, and as with how Microsoft all but destroyed Novell in the distributed server market, many managers in information technology do not have the balls, yes lack the balls, to not pick Microsoft.
About VMware, he notes:
... VMware just costs way to much. This view of mine was reinforced in a recent meeting with VMware, where the discussion of VMware feature set, and the associated pricing became, well, to be fair, enthusiastic to be sure. It was professional, it was honest, and it was quite clear, that VMware was not hearing us. VMware has for the last 5 or 6 years, continued to add features, failed to enhance existing features in reference to scale and scope, for enterprise clients.
In the post, he doesn't weigh in on Oracle, which is rapidly positioning itself to be a major virtualization player. In mid-April, Oracle revealed plans to acquire Sun Microsystems. Last week it announced its intent to purchase Virtual Iron. Even before that, Oracle was setting itself up to be a major player. It is, however, too early to call what exactly that role will be.
My prediction for the the "virtual wars" is the following:
- Citrix will position Xen to focus on the client side. The bulk of its innovation is being found there today. It will likely partner even deeper with Microsoft.
- Hyper-V will be present in the majority of enterprises, but will be the exclusive environment for most SMBs. It may not be free, but it is *part* of Windows Server 2008, and presumably will remain in some form or another in whatever OS follows. Eventually, enterprises need to upgrade, and the initial perception is that it's easier and cheaper to use what's already there. Especially if Microsoft partners to build an ecosystems around Hyper-V, which it already seems to be doing.
- Oracle and VMware will duke it out for the Fortune 500 market, with infrastructure tools becoming the key differentiator. The two companies will focus on the market segment willing to pay for the resources they need, so long as the product actually delivers on what it promises.
If we're using the Unix wars as a model, a fourth prediction comes to mind: Not every player has emerged at this time. Perhaps the biggest impact of the Unix wars was the advent of Linux, which was born of a grad school project in 1991 and began taking off in earnest mid-decade. Eventually, Linux matured to the point of being a Unix replacement in all but the most compute-intensive environments.
Linux, however, is not really Unix, which is to say that the virtualization "killer app" may not be anything like the virtualization environment as we know it.
Amy Newman is the managing editor of ServerWatch. She has been covering virtualization since 2001, and is coauthoring a book about virtualization that is scheduled for publication in October 2009.