Saturday, December 6, 2008

If David Letterman worked for Gartner

If David Letterman worked for Gartner, Wednesday's "Top 10 Disruptive Technologies Affecting the Data Center" keynote from Carl Claunch would probably have been a bit more like engineer/humorist Don McMillan's routine at the Gartner Data Center Conference. Carl's Top 10 list certainly would have been funnier mashed-up with McMillan's good-natured attack on death-by-PowerPoint. But maybe that wasn't the vibe Carl was going for.

Or maybe I've been in Vegas too long.

In any case, Derrick Harris from On-Demand Enterprise gave a good run-down of Carl's keynote (including the Top 10 list itself) and other key presentations from the week. I thought I'd add a few specific comments on the cloud computing and green IT items (#2 and #10, respectively) from what Carl said on stage. And, for good measure, I can't help but add in a few of McMillan's "time-release comedy" nuggets.

How do you know a cloud provider can deliver what they promise? Since you can't spend countless hours asking cloud service providers a sufficiently long string of questions about how they run their production environment for high reliability, you have to fall back on one of two approaches, Claunch said. One is contractual remedies. That one's inadequate. The second (and only currently useful) approach is to base your decision on reputation, track record, and brand. As Claunch said, "there's a leap of faith here. It's early days."

What is a virtual server? The waiter or waitress who never shows up.

(You guessed right. That wasn't in Carl's keynote. McMillan wrote that one especially for us virtualization-immersed conference goers.)

Huge swings in the scale of compute resources needed to support your apps would pretty much force you to try using an external cloud provider. Claunch didn't think organizations will have the ability to support their apps when compute requirements suddenly grow by 10, 100, or 1000x. I'm not sure I agree: it's what companies are having to do now. Of course, it's what leads to massive over-provisioning of servers that end up sitting idle for most of the year. But, if you set up your servers in a pool that's able to be dipped into by any of your apps, automatically provisioned (and de-provisioned when appropriate), all using priorities/policies you set, you actually can do this yourself. Several of the organizations Cassatt is working with now are doing exactly this.

I definitely agreed with his next point, though:

With cloud computing, it's all about service. As Carl said, "All that matters to you is the service boundary." In most respects, you don't really care what combination of hardware, software, or virtual machines is being used to run your applications, so long as they run to your satisfaction. "If they [the cloud provider] can figure out how to do it with rubber bands reliably, that’s fine," Claunch said. "It doesn't matter to you."

What is enterprise storage? McMillan told us, of course, that "enterprise storage" is a closet on Star Trek. May your data centers live long and prosper.

About the Green IT topic: more than a third of the audience would be willing to pay a premium if a new bit of technology is "green." 26% of the people answering Claunch's in-room poll said they'd "buy green" only where it saves money, space, defers data center construction, and the like. That much I expected. What I didn't expect was the 34% who said they'd pay more for technology if it's green. In this economy. Seven percent even said they'd pay a "substantial" green premium.

Finally, statistics are everywhere at conferences like this. Some are intriguing. Some are useless. You can never quite tell immediately which is which, though. (The goofy Gartner game-show music while the voting and tallying is happening certainly doesn't help.) The Gartner analysts, though, helped provide context whenever they did the in-room polling by rattling off how the results had changed from the past few times they'd done similar polls or what they'd heard from end-user inquiries.

In honor of useless statistics, then, I'll leave you with one that McMillan pointed out that left 'em rolling in the aisles. If 44% of marriages end in divorce, that must mean that 56% end in death. That really doesn't seem to leave you with any good options, does it?

And Gartner should breathe a sigh of relief: I think Letterman likes his "day" job.

No comments: