If cloud computing is ever going to reach its full potential, it needs to be more than a utility service that IT managers use to offload excess computing demands. That will require a shift in the way programmers build applications, says Russ Daniels, CTO and V-P of cloud services strategy at HP (s hpq). In an exhaustive interview with Ars Technica, Daniels talks about building applications for the cloud that instead of looking inward for data, can gather it from across various repositories and scale easily.
While we focus on cloud computing primarily as infrastructure, the four-page story goes beyond hardware to build a cloud model that is as much of a utopia as Plato’s Republic. It’s a lovely vision, and many of Daniels’ ideas, such as rethinking the way programs are built so they scale easily and are able to seek out needed data rather than store it internally in the program, are already being implemented. Open APIs help bridge the walls between programs in a way that is seamless to users and research into building scalable software programs abound.
However, the more rhapsodic Daniels gets about the future of the cloud creating this pervasive and persistent access to information from any device, the more I find myself wondering what world he’s living in. He says:
But you want to think about the device differently than simply being this independent tool. You want to think about it in the context of the cloud. When you think about the state that you have on a device, it should primarily be thought of as cache, rather than authoritative. Getting this stuff right takes some top-down thinking …
So when I think about the authoritative data source, it should be in the cloud. And it’s a service, not a repository, so I can deal with the complexities of the real world where there are many potential data sources. However, I have to get every piece of software that’s involved to do the right thing-to delegate responsibility-rather than acting like they’re independent owners and whatever they have in their local state is good enough.
Wow, thinking of your hard drive or EMC (s emc) boxes as cache rather than the authoritative version of your data is a big leap, one that’s going to take more than just top-down thinking. It would also require open standards for sharing data across different clouds, an attention to privacy as well as user identification and authentication, the willingness of businesses and consumers to buy into a subscription model to access data and content, and a business model that rewards sharing data. Oh, and we also need to make sure the clouds themselves, and the network, are reliable.
I like Daniels’ grand vision, but if we assume that data is the most valuable resource a company (or an individual) has, then this utopia is missing a clear business model that will let me take my CardScan data, mix it with my Gmail contacts and then let me access it using the contacts section of my mobile phone seamlessly. Companies want my data to only be accessible to them because it makes them money — either by trapping me as a customer, keeping my eyes attuned to their ads or requiring a subscription or hardware purchase. I hope Daniels addresses some of these issues in the next installment of this story due out next month, because he has imagined the cloud as a way to offer user-friendly convergence — something we desperately need.