Wednesday, June 25, 2008

Heads in the Clouds

Last night I attended Cloud Camp, an impromptu conference in San Francisco that focused on cloud computing. The event was thrown together in three weeks and took advantage of a large number of Web admins, developers, movers and shakers being in town for other shows. This was an unconference, a term coined years back at BarCamp, a collaborative get together that was created to show up O’Reilly’s exclusive Foo Camp. That means there were no scheduled talks or keynotes, only a big paper grid, some sharpies, and lots of enthusiastic folks to talk and ask questions.

When the attendees had announced their proposed sessions and placed them in the grid of times and meeting spaces, the 300 or so attendees filed out and went to chat about what exactly cloud computing is. And the resounding conclusion reached by most was that Cloud is the new SOA. And that’s not a good thing.

The first talk I attended was supposed to be about cloud architecture. Hurrah, I thought, let’s hear about how you open an account with Dell and get those servers into the grid 10 minutes after you unbox them. But, no, the talk ended up being a lengthy product pitch, veiled in a thin smear of “what’s in a cloud stack.” It quickly descended into the leader extolling the benefits of a cloud-based markup language used to describe system stacks. Of course, this was the lead engineer behind said markup language, and it was also the primary product of his startup.

Strike one.

Next, I attended a talk on using Ruby in the cloud, though the talk was ostensibly about reaching 1 billion page-views a month. This discussion focused on the success LinkedIn had using Joyent to host its Facebook application. All I got from this discussion, aside from some excellent Ruby speed tips, was the distinct feeling that I’m missing out on the gold rush taking place inside Facebook applications.

Strike two.

The most interesting part of the evening for me wasn’t the talks, though I hear Google’s Kevin Marks actually managed to spark up a good session, and that Amazon’s Web Services guys were there to listen to complaints. My night was capped off by a lengthy discussion with an unabashed, unashamed venture capitalist. We chatted for a long time about where the money could be made in the cloud. His conclusion was that there would eventually be big roles for middle-men. I called them integrators, but he wasn’t so confident in that term.

Foul tip, just down the third base line.

The trouble with the cloud, right now, is that it’s being used to describe a number of different types of systems. There’s the Google-Amazon system, where you build a non-critical application and host it inside the massive grid of computers at these Web companies. That’s what Cloud is supposed to mean. The other cloud, however, is the internal cloud. It’s a term used to describe a massive grid inside a company, where individual applications are provisioned, allocated, and dynamically resized to take advantage of a slice of this big grid. It’s a commodity in the basement that’s squeezed into injection-molded case scenarios.

Hmmm, sounds an awful lot like service-oriented architecture, doesn’t it? SOA can mean internal systems, connecting and chatting like we always wanted them to, but were never able to accomplish. Or, SOA can mean bringing in SaaS and tools from outside and tying them to internal systems. They’re almost exact opposites. But then, they aren’t at all. They just vie for the same resources, attention and standards. Yet making the Subversion server talk to the change management server is almost entirely unlike making talk to your company’s exchange server.

And yet, they’re very similar. As similar as, say, two clouds. Shapes and forms, speeds and purposes aren’t the real meat of a cloud. The meat is in the viewer. What do you see in that cloud? Oh, Winnie the Pooh! And that one? A rain storm.

If my new VC friend is right, the clouds will soon be filling up with folks who can fill in the mortar between applications, servers and cloud hosts. Not unlike the wildly large ecosystem of SOA tools and products that sprouted up over the last three years, cloud computing will likely become a super buzz word, if it hasn’t already. It’ll be the place where we start to find new standards, new innovations, and new three-letter acronyms.

Let’s just hope that this time, there’s fewer standards involved. The last thing we need right now is a new set of WS*.

-- Alex Handy

No comments: