VMware VMotion Debate: Virtualization Vets Get Passionate, Partisan

Judging by the comments on my last blog entry (on CIO.com and other sites owned by parent company International Data Group in various countries and languages) people are very attached to their own way of approaching virtualization, and the vendors and technology they use to help them out.

(I have to say the comments on CIO.com were a lot more balanced and trenchant than a lot of the other sites, at least as far as Babelfish and I can tell through the language barriers.)

IT pros don’t like it when someone else (even someone with a good track record in VM implementation and management) says some treasured feature is either irrelevant or, more accurately in this case, not as big a deal as they like to think it is.

In this case, the VM manager was Chris Steffen, principal technical architect at Kroll Factual Data, a credit-reporting and financial-information services agency in Loveland, Colo. The commenters’ consensus was that his assertion that dynamic migration is a nice feature to have but far from critical was, at best misguided and, at worst, partisan propaganda or criminal ignorance.

Steffen’s point was that it’s not a good idea to take a VM running a mission-critical application and shift it from one physical server to another in order to keep workloads balanced. That’s not enough benefit to justify the risk of moving a key application while it’s being used.

Stupid, shortsighted, head-in-the-sand denial, according to the commenters (VMware users to a one, as far as I can tell).

The ability to move a VM is not only a critical capability, it’s perhaps the key measure between grade-school-quality, wannabe-virtualization vendors that can’t build live-migration into their Tonka-toy software and real VM(ware) companies who can not only build it, but charge through the nose for it as well.

To be fair, no one denies VMware’s VMotion is a lot more sophisticated at VM provisioning, migration and management than Microsoft’s stuff. Virtual Migration Manager 2008, which is due to ship commercially in September, is perfectly adequate as a migration and provisioning manager, according to Steffen, who’s been testing it forever.

His point had nothing to do with the quality of the migration tools (or their existence, which seems to be a point of debate among VMware partisans). It had to do with the advisability of moving hot servers in the first place.

Which (and this explains why I’m not just complaining about criticism from commenters), is going to be a bigger deal as the VMware-Microsoft competition continues to heat up.

Partisanship leads people to argue over code words instead of the issues that spawned them.

Point out one “on the other hand” about Iraq when talking to someone committed to either a Republican or Democratic position and you risk being accused not only of ignorance, but of willful misinterpretation of the facts based on your own presumed support of the opposite position.

Think it doesn’t happen in IT? Remind a Mac user that the damn things do, too, crash, sometimes so hard you not only have to shut them down, but unplug them and remove the batteries to get them unstuck.

Remind a Linuxian that the amount of time it takes to build and maintain an ethically and ecologically pure open-source server can make it less cost effective than just buy a Windows or Mac machine and run commercial software.

Or, suggest that someone’s favorite virtual-server-management feature isn’t critical, or even advisable, based on policy and service-level issues rather than anything having even remotely to do with technology.

Some of your responses will be based on clear-eyed analysis; others will be based on knee-jerk partisanship.

The problem is that the vendors—who spend a lot of money—have legions of smart, erudite, aggressive spokesweasels fogging up the blogosphere and reinforcing the worst impulses of the knee-jerkers.

As time goes on and the rhetoric gets thicker, it will be a harder and harder slog for IT managers who would rather build and operate their own tech universes than spend all their time fact-checking FUD and counter-FUD from both vendors and other users.

It might do some good for the vendors, and make more work for the analysts, but I doubt it will help the end users very much, no matter how partisan or non-partisan they are.

The problem will be sifting the useful and broadly accurate content from the blowing smoke and FUD, no matter how fact-filled or even how well-intentioned either of them appear to be.