Demystifying The Cloud

It's 2009 and "computing in the clouds" has gotten somewhat mainstreamed over two years since the market entrance of Amazon Web Services put the idea front and center. Yet, like pop diva Pink, it remains largely misunderstood.

Most techies "get" that the cloud is a form of distributed computing. To this, experts add that it relies on a method of virtualization, albeit a specialized type that is expressly set up for multitenancy, that is, designed for consumption by multiple client sites.

But it's in how infrastructure, OS, middleware, and application interact in "a cloud way" (along with who "owns" and "manages" cloud resources) that tends to muddle the concept for the uninitiated.

In a recent report, McKinsey pointed this out, indicating that there were "at least 22 different cloud definitions in common use." Despite this, intrigue is in the air and a bonified source of hype is starting to yield a significant early adopter market.

Infrastructure players like Amazon can deliver services via the cloud, known as infrastructure-as-a-service (IaaS). Platform providers like Microsoft's Azure (which works with .NET) or Google (via Google Apps) can deliver platform-as-a-service (PaaS), which can be thought of as customizable frameworks, from which end-users build applications. Finally, there are service-based clouds that deliver applications, such as SalesForce.com for end-users that don't require customization.

Gartner Fellow David Mitchell Smith reminded CIOZone that supplier definitions often depend on their place in the ecosystem. So, little surprise then, when infrastructure players talk about "abstracted infrastructure that can host end-user applications," says Mitchell Smith, while platform as service providers might emphasize "services that yield applications that can do more or handle data in a more efficient way."

Originally, the cloud simply referred the internet itself, a hinterland of interconnection outside your firewall used by everybody and only partially controlled-and not without persistent corporate elbow grease and a platoon of security providers. And to Hewlett Packard's Chief Technology Officer, Russ Daniels, that's where the cloud computing definition should begin. In a recent talk at the Cloud Computing Conference, he described it as "the next stage in the evolution of the internet."

Daniels went on to say the cloud is a model where there is "programmable access to memory across the internet in a way that adds value in terms of solving specific computing problems." [On June 4, Palo Alto-based HP announced that it was collaborating with Verizon Business, a unit of Verizon Communications, to bring to market the company's new Computing as a Service (CaaS) for data center operations management and monitoring.]

The CTO said that HP sees the cloud as particularly good at establishing connection among partners in an ecosystem. He views the distributed, easy to utilize nature of cloud as a culmination of what has gone on with services oriented architecture (SOA), where discrete components combined on demand to get work done and the data stayed separate from the application; and Web 2.0, which added rich user experience to applications via AJAX and a social connotation to computing.

As heavily reported in the blogosphere late last summer, start-ups like Animoto relied on cloud-based infrastructure from Amazon to scale large (when FaceBook featured them) and smaller (when the fad seekers dropped off, leaving behind a smaller group of permanent users,) recounts Bernard Golden, chief executive officer of HyperStratus, a San Carlos, Calif.-based consultancy specializing in cloud services development. But you don't have to new and tiny to benefit. New York Times became another high profile example among those who know the cloud back in 2007 when they "borrowed" infrastructure muscle from Amazon to convert and store historical images.

To Gartner's Mitchell Smith, the notion of elasticity inherent in doing things "in a cloud way", the idea that you run at large capacity (like scale) but automatically revert to smaller consumption of infrastructure, is a key distinction. "This elasticity of getting big and small, is really what sets the cloud concept apart from other types of distributed computing," the Gartner Fellow says.

At the recent InterOp conference, a panel discussion of "cloud makers"-including Strangeloop, Hyperic, and GoGrid- focused on the elasticity concept, "What Elastic Capacity means for IT Operations," explaining the flexibility the cloud can offer. David Link, CEO of ScienceLogic, did a writeup of the event on his blog in which he admitted that many enterprise IT conference goers seemed more cautiously interested in learning more than already steeped in familiarity based on projects at home.

Today, experts guestimate that less than 10% of computing is handled by the cloud, but Golden thinks the model will steadily become more popular over the next five years. IDC expects IT spending on cloud-based services to grow threefold, reaching $42 billion by 2012. Early work has tended to focus on access to more affordable infrastructure and applications that have variable use patterns, notes Golden. "Our clients are trying to get beyond the broad concepts and get started on projects," he says of his work.

He discussed a project done for Silicon Valley Education Foundation, which uses a peer learning collaborative tool to file and share lesson plans and is used by 13,500 teachers and 260,000 students throughout Santa Clara County. A cloud based approach was adopted to set up the equivalent of co-location for continuity. "They got full redundancy for 50% of the cost of the alternative backup solution," Golden says.

Stuart Charlton, chief software architect for Elastra, San Francisco says the cloud-based computing market is developing for use cases that include applications for business intelligence analysis of large data sets. (As widely reported, Pfizer, Eli Lilly & Co., Johnson & Johnson and Genetech are among the pharmaceutical companies that are piloting various use cases, including data analysis.)

In the case of Elastra, capabilities delivered in the cloud support the mapping (and automating the transfer of) computing environments as part of the server consolidation process. "So, what needed to be done before in several steps, with plenty of manual work involved can be done much more efficiently," he explains.

"Nearly every Fortune 500 company that we've communicated with is kicking the tires and considering how cloud-based projects could work for them," says Charlton. "Pilots have gone on throughout the recession," he adds. "I wouldn't say there's been a huge increase in demand, but we think this autumn will be a busy time for us."

And tech heavyweights, including HP, IBM (with Blue Cloud), Microsoft, and Sun, with its Project Kenai of APIs for the Sun Cloud Service and Open Cloud Platform, are all working studiously on cloud initiatives, even as their education efforts expand to clear up any remaining questions about doing things "the cloud way."

Besides, advocates say, the fact that computing in the cloud is vague is in keeping with other great computing trends that proved to be innovative. This includes client server computing in the late 1980s and object oriented computing in the early 1990s, says Charlton.

	1. 06-08-2009 12:33
	A relevant facet of cloud computing adoption that in my opinion doesn\'t get much air time is that it inherently involves embracing a proprietary platform in order to achieve the nimbleness or elasticity benefits, as each provider\'s API is different and therefore represents friction if/when changing providers. We ultimately face this proprietary issue in every facet of computing, although I think the romantic notion of the cloud is that it is plug and play...the transition costs might be less than traditional means in switching from one provider to another, but it is more accurately described as \'plug and play with concerted effort\'. Registered

Frederick B. Kauber