Wednesday, November 11, 2009

Two cloud computing Rorschach tests: 'legacy clouds' and the lock-in lesson

This week's San Francisco Cloud Computing Club gathering was a great place to meet some of the movers and shakers in the cloud computing market (or at least the ones within a short drive of San Francisco). The event's concept was to spend some quality time talking through cloud computing issues with a crowd of people who spend all day thinking about the cloud and working on making it a reality.

James Watters (now at VMware) and crew invited an intriguing crowd of about 35 or so (here’s a Twitter list of attendees). Gary Orenstein snapped some pics. John Furrier (from siliconANGLE) used some magical little device to record the entire multi-hour discussion for posterity. Hopefully the recording bit will turn out to be a good idea. There were a few pithy quotes, for sure.

Raising 2 key questions about cloud computing

As James Urquhart (CNET blogger from Cisco) guided the conversation, I heard two interesting points of contention that divided the room pretty dramatically. It struck me that each comment could actually serve as a Rorschach test of a person's view of how powerful and important the concept of cloud computing actually is.

One question was about where IT has been; the other helps define where we might be able to go. Both say a lot about the approach to cloud computing that you're likely to take.

#1: Is cloud computing compelling enough to change how IT runs existing apps?

I lobbed a question for the group to chew on that went something like this: we're hearing lots about new, pre-integrated stacks of hardware, software, storage, etc., that are intended to be used to create private clouds. These systems are interesting for new, greenfield applications. However, even if these sell like hotcakes, most of an end user organization's IT systems are going to still be using what already exists in their data centers.

The question: is it worth the pain and suffering to try to shift what's already running -- and contributing to your business -- to a private cloud? Should you create a "legacy cloud" from what already exists?

History -- and IT conventional wisdom -- say it's too hard to mess with things that work. If it ain't broke, don't fix it. The disruption of changing things adds IT operations risk, and the IT ops guys are not fans of that. Optimists, however, say that cloud computing (even if it's of the internal, private kind) is too compelling to pass up, and IT ops should make the leap.
One person in the Cloud Club discussion compared it to having to till the land to grow the wheat to make bread, versus suddenly having access to a supermarket that sells ready-made bread. You'd always choose the supermarket, right? Not so, if you're in IT operations.

I side with the optimists, but am realistic about how you'll need to get there. Customer feedback continues to say that a Big Bang approach (send everything to a cloud -- now!) certainly won't work in large companies. On the other hand, an incremental shift, app by app, is a lot more palatable by organizations with large, complex IT systems. And, in the end, there will be some apps that never move. (Last week's post mentioned John Treadway of Unisys -- he did a good presentation on a methodical, incremental approach at Cloud Computing Expo in Santa Clara.)

By the way, the answer to this question is not insignificant: it will determine how large the private cloud market will end up being. And, actually, how large the public cloud market will be as well. If end users are too timid, or unconvinced of cloud computing's benefits, this revolutionary change ends up being an uninteresting blip in the IT timeline.

However, my bet is that this process picks up steam bit by bit and is hard to stop.
#2: Has the industry learned its lesson about lock-in?
The second telling question about cloud computing I heard this week is as old as the industry itself, but with a new, cloudy twist. It's a follow-up to the previous questions in many ways. Tom Bittman of Gartner and many others have talked about the private cloud market as just the opening act for cloud computing. At Cloud Club, we talked about private clouds setting the stage for the ability to move some or all of your computing to an external cloud service somewhere. Making platforms slippery enough to be able to handle the shifting back and forth of workloads led to a discussion about how exactly that can come to be.

And the thing that might keep it from being: vendor lock-in.
One potential future tossed around at Cloud Club by James Urquhart, Rich Miller (of Replicate Technologies), and Tom Mornini (of Engine Yard) was that the eventual normal computing scenario will be one in which there are sets of interoperable clouds that customers will be able to choose from and move between, rapidly and easily. This Intercloud concept, as the Cisco folks call it, requires easy interoperability.

And, it requires that the industry (users and vendors alike) have learned the hard lessons about lock-in. The optimists and advanced technologists claim that we already have (Tom of Engine Yard certainly falls into that camp). If it's true, you'll have a slippery infrastructure supporting applications wherever it is that they need to run.

But skeptics didn't let that pass by without a fight. The concept seems too good to be true, and contradicts the way computing has progressed so far.

The economics of the vendors in many ways work against making it easy to move workloads around. Besides, big end user companies are used to making bets on technology stacks with only a few vendors. Is that no longer needed? Customers would certainly prefer the economics that result from making this possible. Will vendors cede control because they know lock-in is a losing proposition? Have we learned the lesson of lock-in at last?
You can see why this discussion needed comfy chairs, a moderator, and a healthy supply of beer and wine.
And the answer, please…
Obviously, there is no answer yet. There's not enough data one way or the other. The market, as we all know, is just getting going. But don't let that put you off. Why?

I think where you stand on this point is actually pretty intriguing: one position is a pragmatic read of what's come before; the other is an optimistic assessment of what's possible. This is a test of whether you see cloud computing as a fundamental shift or just a new buzzword to be exploited by those providing the old thing.

Of course, there is the possibility that cloud computing changes the game, but those who liked the old rules will continue to try to enforce them (read: making money from lock-in). The newcomers will try to undercut those rules and redefine the game. Innovator's Dilemma anyone?

So, my point in highlighting this subset of the Cloud Club conversations was this: keep an eye on these questions. They are some of more interesting things playing out in the cloud computing space.
And, for as much fun as it was to spend several hours in good-natured arguments over these and other topics in a penthouse in San Francisco, none of the answers will appear magically, nor overnight. We leave that to time, actual customers, and the market to decide.


James Watters said...

"This is a test of whether you see cloud computing as a fundamental shift or just a new buzzword to be exploited by those providing the old thing."

Great quote...

I think the best thing we can do at CCSF is tease out some of the things that are very 'real'...its pretty hard to 'hype' that room.

One of Tom's cool themes was developer productivity--and I think that no matter how long it takes cloud can be a massive boost to developer productivity (faster dev, test, deploy, scale, decom)..I think that is one theme that can shape the future of IT deployments even if it can't touch the past..

Jay Fry said...

Yep, it's definitely a tough crowd. No hype zone for sure. :)

I think you're right, developer productivity is certainly one thing that the cloud can improve pretty easily. I'm betting companies trying it out are also likely to say, "Hey, I should also use this for...(fill in the blank)." Seeing things work can be pretty persuasive, even on skeptics (and there are a few for cloud computing).