Sunday, March 18, 2012

Visiting Cloud Connect 2012 Santa Clara

Last month I went down to the Cloud Connect 2012 conference in Santa Clara.  If you're wondering why it took me a month to write up a synopsis, it's because cloud computing is pretty far afield from the business topics I usually cover on this blog.  I had to take the time to understand just what all this cloud stuff is about.  I'm not some tech-addicted early adopter; I prefer to see how new developments shake out before I think about jumping in.

I'm all about keynote speakers, and there were plenty of them on the platform.  Steve Wylie, Cloud Connect's general manager, set the stage for us.  Hey, what happened to the gal who promotes other UBM events?  She must be busy.  Anyway, Steve briefly mentioned that cloud's main effort has shifted from defining its existence to building a practice.  I guess we can look forward to more intermittently employed IT professionals calling themselves cloud consultants.

First up was Alistair Croll, currently with Bitcurrent.  He described the evolution of measurement from feet to horsepower and later conventions to argue that computing power as a unit of measurement is becoming obsolete.  IT professionals must add value with virtualizations because infrastructure is moving to IaaS: Infrastructure as a Service.  That was the first of several terms I learned at this conference, all of which seem hard to define because they bleed into each other.  Like, where does IaaS end and PaaS (Platform as a Service) begin?  Hire a cloud consultant and they'll explain it all to you.

Lew Tucker from Cisco told us that clouds come in public, private, and hybrid flavors.  The cloud model that an IT practitioner prescribes for a client has to fit.  I saw some similarities in Lew's design perspective with what Alistair presented.  I grabbed Lew's thoughts as fast as I could:  IT pros build apps on several layers of abstraction; open source development helps build clouds; having a "network service" provides linkages as a translation layer between complex physical infrastructure and user experience; controlling host names and IP addresses is the key when creating virtual isolated networks in a multi-tenant environment.

Bill Gillis from Beth Israel Deaconess Medical Center said that losing paper charts would be a critical problem in patient care, so building system access was a key thing.  Building a private cloud maintained the center's control over patient information (aha, IT pros, your private clouds are attractive for a sector very focused on the confidentiality of sensitive information).  Government stimulus money has pushed private practice physicians to migrate patient records to the cloud.  Hmmm, I wonder if that's a Medicare mandate now.  Anyway, Bill also said that many hosted apps have known security flaws, so virtual patching can apply fixes.  If you really must know, virtual patching establishes rules for a firewall that don't alter the underlying code.

Jan Jackman from IBM was up next.  BTW, IBM is all over the cloud space and I suspect their hidden agenda in sending representatives to cloud events is to find technology they may want to acquire.  That's what I'd do if I were them but I have no way of knowing whether they consider any of the vendors here to be acquisition targets or potential JV partners.  Anyway, Jan said there's no one way to view monetization.  Content providers can pay for only those parts of the cloud service they actually use.  IBM uses an open source ecosystem of many partners.  Multiple ecosystem partners participate in every step of the cloud development lifecycle model.  I wonder whether adding too many partners inhibits topline revenue growth for a cloud value chain.  I mean, at some point your number of partners becomes unmanageable and monetizing them all may not be worth the effort if they don't all deliver.

Geva Perry, author of Thinking Out Cloud, said CIO magazine surveys from 2009-2011 showed a gradual adoption of cloud computing.  Organizational "creep-up" showed that adoption started at the bottom with developers.  The phrase "CIO is the last to know" shows how top IT executives don't drive cloud adoption. IMHO the irony there is that CIOs are the ones who ultimately determine capital spending budgets, so the legacy architectures they approved must have had sufficient bandwidth and internal flexibility to enable the proliferation of a cloud ecosystem.  Anyway, Geva mentioned the federal government's own apps store, which I didn't know existed.  Geva said vendors should design and build products that enable viral adoption by the rank and file people in an enterprise.  IMHO, the lesson for corporate honchos is that small, nimble cloud apps will move faster than a big IT project that has to go all the way up to the CIO for approval.

Jesse Robbins from Opscode was my favorite speaker at the entire conference.  The brilliant hits just kept on coming from this dude.  I'm going to restate his best points without interpretation.  He said every company learns that complexity increases with scale and that latency can cause many failures.  Designing for failure allows resilience.  Large-scale fault injection across multiple systems is called "resilience engineering," putting people first.  IT pros should expect system failures and plan periodic scenarios that anticipate failures.  Physical safety standards and building codes are products of what structural engineers have learned about building failures.  Applying this to IT means living through failures can help design improvements.  Cloud pros shouldn't rely on one-time load balancing; instead they should build incident management tools around routine daily processes.  "Infrastructure as code" gives you many knobs and levers to control systems.  Mean time to recovery (MTTR) is a better metric than mean time between failure (MTBF).

Alistair Croll made a repeat appearance.  He said that up to now content creators have made their money from scarcity but consumers want content immediately.  Merging content with software and devices is messy for providers but consumers like to copy and tweak.  I had to reflect on this from my perspective as an amateur user of web content.  IMHO the demand side will eventually win unless bad laws (SOPA!) interfere with the market's natural evolution.  I'll never forget a mantra I heard in the 1990s during the Internet's early days:  "Information wants to be free."  Just as I was beginning to digest this stuff, Alistair threw out a curve ball by mentioning the Thingiverse.  OMG, it's a whole mess of open source designs for 3D objects!  That is exactly the kind of resource a System D entrepreneur could use in desktop manufacturing!  Anyway, I'm pretty sure Alistair mentioned it as an example of an open source approach to growing cloud architectures.  Alistair finished by saying that stepping away from "root" ability to hack/open/meddle in content will be an attack on general purpose computing and toward monopoly.  Users and developers should fight for the freedom of computing.

I missed a couple of early Wednesday speakers but the panel was cool.  The panel's consensus was that open source architecture is the way forward.  Architectures face a choice between open/complex versus closed/simple, so don't get trapped in one that's not adaptable.  Service providers with closed models will eventually see the benefits of breaking down firewalls to allow outside apps.

Zorawar "Biri" Singh from HP said "hybrid delivery" spans three areas:  non-premise private cloud; managed cloud (i.e., a delivered service for a customer's specific need); and public cloud.  I suspect this taxonomy is HP's way of branding its service offerings across the cloud spectrum.  Anyway, Biri gave us a pretty good definition of "open stack" as information layered upon apps that are in turn layered upon infrastructure.  That's as good enough of a rule for me as I try to sort through SaaS, PaaS, and IaaS.

The final keynoter I cared to hear was Robert Holleyman from the Business Software Alliance.  He thinks Japan has the best regulatory model and that Europe is an excellent export market for U.S. cloud providers. IMHO we won't be exporting much cloud stuff if we have to rely on Indian and Chinese IT pros who are increasingly returning to their home countries after their education and initial training here.  Anyway, Robert thinks that Brazil and China will be large, growing cloud markets but may not be open to the U.S.  That makes sense; why should they be open to us if they have grown their own IT talent for a generation?  Robert emphasized three things the U.S. needs to compete effectively:  consistent privacy policies; government buying power used constructively on the best technologies; and prevention of cybercrime and IT theft.  I say good luck getting any of those from government agencies that will soon be starved for funding.  If the cloud industry wants coherent policy and funding from Washington, they need to lobby for it and find those lawmakers who truly "get it" outside of Silicon Valley.

These keynotes filled up the mornings so I had to find something to do in the afternoon for the two days I had a free expo floor pass.  The booth people were cool but frankly their explanations of what their respective companies do went over my head.  Microsoft had one of the coolest booth configurations.  Some other purveyor of cloudiness brought in a 1950s-style aluminum camping trailer.  I learned a few of the basics of this emerging subset of computing but I'll leave the heavy lifting to the hard-core IT community.  I did score some free T-shirts from a few vendors.  I was also grateful that a few vendors staffed their booths with some attractive women.  That's reason enough for me to return to Cloud Connect next year.