How Semantics Can Help Cloud Computing

By SemanticWeb.com on February 19, 2009 1:20 PM

Jennifer Zaino
SemanticWeb.com Contributor

Semantic web, meet the cloud.

But the multi-cloud computing world we’re moving to needs to unleash itself from the host of application programming interfaces
(APIs) used by each new provider, one expert says.

This trend is bringing the industry down the road of vendor lock-in that might ultimately prevent users from choosing a new cloud-computing provider based on location or other features that might also be important to them.

At a high level, today’s cloud platform and infrastructure providers are doing much of the same thing, but require applications to interact with them in completely different ways.

“They use completely different terms and API structures, so if I want to move my application or have a combination of both, I have to re-architect my application for both APIs,” says Reuven Cohen, the founder of Enomaly. Enomaly provides the Elastic Computing Platform programmable virtual cloud computing infrastructure, which is utilized by hosting companies that want to build clouds. “That’s difficult. I don’t want to re-architect for every new provider of capacity out there.”

The problem is the same for enterprises that might use a VMware infrastructure internally and only occasionally want to turn to the cloud for additional external resources. “Do I want to re-architect my entire application for that one day in a month where I have a sudden spike in traffic?” Cohen says.

Unlikely, and the semantic web offers a solution to the emerging problem. It starts with work being done at the Cloud Computing Interoperability Forum, to create a common cloud taxonomy and ontology, which is a way to express cloud computing and its subsequent parts in terms of a consensus data model.

“It’s nice to create a data model, but it’s nicer to do something with it,” says Cohen. “Semantics provides that opportunity,” and Cohen is in fact leading a group at CCIF to apply the Semantic Web to APIs as part of a broader effort to create a unified cloud interface — a Semantic Abstraction Layer, single programmable interface for all other APIs, with OWL (Web Ontology Language) serving as the basis of that model.

If successful, it potentially opens up the door to new opportunities in the form of a global cloud and what it can enable. Imagine scenarios where enterprises can, for example, do global performance testing to see how applications respond based on using a series of clouds connected together for simulations of 5 million users,” he says. “There’s a whole variety of applications enabled by this unified cloud.” Cloud vendors who might like the idea of locking in cloud customers are missing the bigger picture — a unified cloud interface isn’t so much a threat to send customers into the arms of competitors as much as it is a chance to give cloud providers more weight with users because of what they can enable.

Cohen posits that work like his might be the real killer app for the semantic web.

“The problem with the semantic web is that no one has figured out an optimal use case for it,” he says. “What I think the killer application for the semantic web is in the ability for machines to interact with one another in a common way, and looking at it from the point of knowledge management is the wrong way. The better opportunity is in the management of machines and how they interact on a broader basis.”

So far, interest in what the CCIF has underway seems promising on the vendor front, with Cohen pointing to all the major vendors showing up at the first event the CCIF held in September last year. In April he expects the announcement of a preliminary implementation of an ontology and taxonomy and possibly even of a semantic abstraction layer.

Comments