Wednesday, March 16, 2011

Real math and rapid-fire ideas: overheard at Cloud Connect 2011

I spent several days last week at Cloud Connect in Santa Clara and I am very glad that I did. It was great connecting with a group of the cloud computing whiz kids (and some only slightly older than that). Plus, Alistair Croll and the team proved that they can continue to put on a conference each year that feels the pulse of the market and provides the content that needs to get airtime.

Last year’s event was solid; this was even better.

Here were some of highlights and things I overheard while I was at Cloud Connect:

The rapid-fire, 10-minute keynotes were great conversation starters. I think they planned it that way. And it worked. The keynotes I saw ranged from Amazon CTO Werner Vogels talking about how the lines between IaaS, PaaS, and SaaS were blurring (thanks to some William Vambenepe-inspired graphics)…to the Eventbrite speaker talking about real-world cloud operations experiences at his current company, at Digg, and Playboy…to Berkeley researcher Oriol Vinyals showing how teaching computers to play and learn complex video games like Starcraft could have great possibilities in cloud resource management and automation. (From my Cassatt experience, this kind of automation is powerful – and hard.) There were only a few sales pitches we had to sit through (Rackspace, this means you), but at least they were short. The quick, concise keynotes were a great way to stretch your mind.

Lots of math. Real math. From the Cloudonomics track to the keynote by eBay, there were some real, hard numbers presented. That’s a home run for anyone interested in cloud computing. The eBay keynoter talked through the math they used to decide if and when to use internal infrastructure and if and when to use public cloud capabilities -- and what situations caused those decisions to shift. HP’s Joe Weinman walked his session attendees through the logic behind his proofs of the inevitability of cloud computing, the situations where it makes mathematical sense, and the reason hybrid clouds are just a good idea. Joe’s work is both thorough and practical: he describes how to use logic and economics to make cloud decisions. You can check out Joe’s work here; he actively solicited people to continue to pound on his numbers and check his work. That’s refreshing.

Does cost still trump everything? The one thing that Joe’s sessions really got me thinking about was this: how far can you take any mathematical conclusions about when an organization should adopt cloud? Humans, after all, don’t always decide things by the numbers (yes, even in IT). Funny thing is that I think it was Joe Weinman who said in his session, “There’s no such thing as a rational economic decision maker.” We can’t look at just cost, he said. “Things are more complex than that.”

Nevertheless, you have to have the numbers. It’s important to have those numbers to refer to, even if we don’t use them as the sole decision-making criteria. Panelist Ravi Rajagopal (from CA Technologies and a lecturer at NYU on cloud computing) said he believed that after all is said and done, customers must use cost to justify any move to the cloud. Speed, competitive differentiation, and other criteria are nice, but they take a back seat. The other panelists (James Staten from Forrester, especially) underscored the power of agility as a driver and the need to look at softer things, like opportunity costs. I personally think we’ll be able to move away from a purely cost-based justification as things like the Service Measurement Index for cloud services take off.

What you get when you mix cloud and accounting. At Cloud Connect I heard a comment that made me think that there may be some interesting accounting tricks still to be learned about cloud computing. One of the big pitches for cloud is that it moves many IT expenses from being capital expenses to operational expenses. However, in one of his Cloudonomics sessions, Weinman noted that Netflix is using AWS reserved instances and amortizing the costs over time. It got picked up and discussed in a bit of depth by Reuven Cohen, Data Center Knowledge, others. An interesting concept, for sure, and one that I hope those with more of a finance background than I will delve into.

What you get when you mix cloud and lawyers. Dr. Chenxi Wang from Forrester did a surprisingly thorough overview of the global legal issues that cloud is causing and that must be addressed. (I say surprising because she was actually filling in for another analyst at the last minute.) She tossed around some interesting concepts and ideas that were new to me. In talking about cross-border data flow, she mentioned an intriguing idea that might help solve some of the legal issues: treating data centers as “data embassies” where the laws match those of the country of the source of the data, rather than the location of the physical data center.

Also, she mentioned that one of the biggest concerns about housing data in the U.S. is the Patriot Act and the risks to the control of that data. But just putting a data center in Europe doesn’t guarantee that a U.S. company is free and clear. “Think the Patriot Act doesn’t impact your EU data centers? Think again,” said Dr. Wang. The legal precedents aren’t yet clear, and are heavily impacted by how you set up your European organizations.

Having your booth next to the beer is never a bad idea. Sure, the CA 3Tera AppLogic booth was in the back corner this time around, but somebody was looking out for us when they planned the location of the liquid refreshments. Having our new service provider partner ScaleMatrix join us in our booth was a great way to immediately share real-world experiences with booth visitors from the get-go. And that freed up CA staff like our man Justin to work on his mad 3Tera demo skills on camera.

Finally, here are a few quotable quotes I scribbled down throughout the conference. I thought these were pretty good summaries of a few of the key topics:

James Staten, Forrester Research: "The power of cloud computing comes through these two words: ‘down’ and ‘off.’" Concise…and true. When you turn a cloud application or service off, the billing stops. It’s straightforward to expand when needed, but cloud computing has to contract as well.

Randy Bias, Cloudscaling: "Cloud is a complete re-think of the IT stack." In other words, the value of the applications are what's important, not infrastructure ownership.

Ravi Rajagopal, CA Technologies & NYU: “We need to bring rigor to change the ‘art of cloud computing’ to the ‘science of cloud computing.’” And, as noted above, “You can’t justify cloud without the cost making sense.” At least, not yet. But we’re getting there.

Joe Weinman, HP: “In most cases, hybrid clouds are cost-optimal as well as the most pragmatic.” He then followed that with lots of small fonts and detailed proofs. And that’s a good thing.

Werner Vogels, Amazon: “It is still Day One in cloud.” Meaning, there is a lot of innovation to yet to come.

If that doesn’t give you your fill of Cloud Connect info, I’d suggest a couple other really good write-ups that I read: check out what Krishnan Subramanian wrote at CloudAve and Bernard Golden’s commentary at And start making your travel plans for next year’s event.

No comments: