So I've got this theory. I believe I've mentioned it here before. The theory holds that Microsoft's stranglehol- …er, um, dominance in the corporate desktop OS market is largely due to Office and its market share.
This sprang to mind the other day as I was waiting in line at the local Starbucks for my beat-me-with-a-stick $4 cup of coffee. Next to me was a surly fellow working feverishly on a Sony Vaio notebook. This notebook had a number of Sun Solaris logos plastered to its cover. Turns out he was an "OS Ambassador" for Sun and was putting the finishing touches on a presentation about the flexibility and advantages of Solaris.
Being an unscrupulous journalist, I simply had to glance at the screen to see whether he was actually running Solaris. And he was working in Windows XP, using PowerPoint. Of course, he had the air of a marketing guy, and getting one of those folks to let go of PowerPoint in favor of OpenOffice probably requires a handgun and electrified body clamps.
That said, it's not like Unix isn't making its own waves. After all, Steve Ballmer -- aka the Anti-Penguin -- recently had to drag his reluctant posterior to the podium at the Microsoft Management Summit 2005 in Sin City and demonstrate Red Hat working under Microsoft Virtual Server 2005. Something the company had been deliberately trying to avoid.
Then again, it seems like the whole world is looking to interoperate itself into one giant bubbling technical glob. Apple wants to run on Intel, and the world is suddenly talking about dual-boot machines like they were the bee's knees -- as though no one ever ran Windows on a Mac using VMware before.
And it seems as if everything from the corporate desktop to a Sony PlayStation wants to run in a grid. Ever since folks started building redundant arrays of inexpensive disks, hardware and OS manufacturers have been looking for new ways to raid the concept of grid computing. Even RAID itself is now being "virtualized."
No, I'm not kidding. Already a virtualization concept, RAID is being pushed one step further, divorcing itself from a pure hardware concept to something much more broadly applicable, namely a software-based object-oriented standard. This is where the concept of RAID is dropped all the way down to the file level. It means administrators can decide what's covered by redundancy and what isn't -- even elements in the same file. Some vendors, such as EMC, already have RAID at the file level working to some degree, though nothing you could call standards-based.
Additionally, there's another set of technicians who want to take the inexpensive out of the RAID acronym. Instead of building RAID systems with bushels of disks slapped into semi-intelligent chassis, these folks are stringing together multiple RAID-equipped computing nodes to form RAIN: redundant arrays of independent nodes. While it's not pouring, raining, or even drizzling such solutions yet, the market can certainly see uses for what amounts to large server farms dedicated to the function of storage. That's what a RAIN node is, after all: a string of many servers, each with independent disk, CPU, and RAID capabilities all unified by a single management platform aimed at truly high-performance storage.
As a physical example, consider a ClearCube blade-workstation array. Each workstation blade has an 80GB hard disk, and the ClearCube management software tracks the empty space on each blade's disk. It then ties all that empty space together into a virtual RAID array spread out across the entire workstation grid. Now instead of each blade having a single disk, imagine if every workstation blade had its own RAID subsystem. Man, you'd have to hire consultants if you wanted to crash that system.
Definitely high-performance -- but just as definitely not inexpensive. And also nowhere near a RAIN standard, partially because RAID itself was never really a standard, merely an architecture. But it certainly gives one pause for the future of the average server farm. And maybe, just maybe, it might do away with some of this obsession about operating systems.