The Uberating System
September 22, 2008 Timothy Prickett Morgan
“Ah, but a man’s reach should exceed his grasp, or what’s a heaven for?” A good question, posed by English poet Robert Browning. I don’t know if the data center can qualify as a kind of heaven, though I do suspect that if there is a heaven, the temperature is always perfect, and I can imagine the faint humming of the workings of the Universe being audible. Anyway, you can’t blame IT vendors for having their reach exceed their grasp, which happens all the time. This is, apparently, how progress is made–through stretching. Sometimes the truth, sometimes our minds. Sometimes both.
Moreover, no one would ever accuse IT vendors of having a grasp that exceeds their nerve, as Nikola Tesla was quoted as saying in the movie, The Prestige. It is unclear if Tesla, a supergenius who was also a bit of a mad scientist, actually said this. But it is not hard to imagine Tesla saying it, and perhaps spitting on the ground and shaking his head. If Tesla were alive today, or rather a few decades ago, he would have, unlike Al Gore, actually invented the Internet. And Linux. (Well, Nicolux.) And probably Web 4.0, whatever on earth that might be (but surely is a future buzzword).
With the dizzying complexity in data centers these days, everyone is trying to figure out how to make IT less complex. A set of announcements from X64 server virtualization juggernaut VMware last week made me chuckle, and I heard echoes of past over-reachings and under-graspings inside of my head.
VMware was hosting its VMworld 2008 show last week, and it was a coming out party of sorts for Paul Maritz, the former Microsoft hot-shot who replaced VMware founder Diane Greene in July as president and chief executive officer. Maritz was part of the development tool team at Intel before he took a job at Microsoft in 1986, where he was part of the five-person senior executive management team until he left in 2000. During his tenure at Microsoft, Maritz was responsible for steering the launch of multiple releases of Windows 95 and Windows NT, the SQL Server database, and various development tools. After his departure, Maritz set up a company called Pi Corporation, which created a cloud-based personal information management system. (Everything is soon going to be “cloud-based,” so get used to it.) In acquiring Pi, EMC established a Cloud Infrastructure and Services Division and installed Maritz as its president and general manager.
Given his background, and the desire by VMware to put an embarrassing mid-August patch bug that caused the ESX Server hypervisor at the heart of VMware’s Infrastructure suite of virtualization tools to shut down and hibernate, you’d expect VMware and its new president and CEO to come out a-reaching not just for the stars, but also for IT budgets. And VMware’s Virtual Data Center Operating System, or VDC-OS for short, did not disappoint.
You have heard a lot of this stuff before; IBM called it On Demand Computing, Hewlett-Packard called it Adaptive Infrastructure, and Sun Microsystems‘ was still working on a clever name a few years back when its business hit a wall and it started to stick to its server and operating system knitting. In any event, VDC-OS has all the important elements of buzzword bingo–it is virtualization, and it is a cloud computing enabler, and it manages networks, servers, and storage all seamlessly and transparently, and it provides fault tolerance and disaster recovery, and . . . .
And does anyone believe for a second that any one set of products can do this complex work in a real data center?
I applaud the notion of a data center operating system, and I said so when Scalent Systems launched its own riff on the concept of a data center operating system, called Virtual Operating Environment, or V/OE for short. (The slash really makes it an OS, you see–something that IBM has forgotten.) V/OE 1.0 virtualizes servers, storage, and I/O and initially supported Linux and Windows servers, and then it eventually supported Solaris and AIX servers. And that is pretty good coverage. But, like VMware’s VDC-OS, coverage is limited.
Whenever I see such announcements, I never see anyone say this:
“Our product is not just about managing Linux and Windows, whether physical or virtual. But rather managing all kinds of servers in the installed base, and that includes AS/400s and mainframes, obscure open source Unix and Linux platforms, various middleware and database stacks, and the widest variety of storage and networking as is humanly possible. And the reason why is we understand that for a data center-level operating system to work–whether it is virtual or not, and perhaps best called an uberating system–it has to be broad and deep in its coverage. We are not asking you to change your platforms, but rather facilitating your desire to keep your platforms and manage them better. And that is why we are dedicating more of our budget to provisioning, patching, updating, monitoring, and managing for a large number of platforms and less to executive salaries and the corporate jet.”
VMware is really a Windows product at this point, with about 90 percent of ESX Server shipments estimated to be on Windows platforms, which have been sorely in need of virtualization. And now that Microsoft has the freebie Hyper-V hypervisor out the door for Windows Server 2008, you know that VMware has to do everything in its power to change the subject. Which is how you get an incomplete set of announcements like the VDC-OS.
I am not suggesting that VMware not do the VDC-OS. An idiot can see that the server has been a commodity for a decade, and the hypervisor is going to be a commodity even before it is widely used in the data center–which is a neat trick, when you think of it. And I think that VMware can and should extend its VirtualCenter and related Infrastructure management tools so it can manage as many platforms as its exuberant $12.4 billion market capitalization suggests it can afford to cover. But that means admitting that ESX Server is not the be-all, end-all, and that X64 servers are not the only machines that matter. And I think Maritz and his team, like Greene and her team (which included herco-founder, husband, and chief technology officer Mendel Rosenblum), have no interest in such a move. In fact, it seems to be beyond their comprehension.
Which means there is an opportunity waiting, just out of reach, for someone to try to grasp.