ComputerWorld.in

Interview: Joe Weinman, AT&T, explains cloud computing

Joe Weinman, Strategy and Business Development VP, AT&T, has been following the cloud phenomenon closely. He practices what he preaches – he is a proud adopter, a vociferous proponent, and he even has a blog dedicated to it, called cloudonomics.

Computerworld.in got a chance to catch up with Weinman, during his recent visit to India. Here are excerpts from the interview:

On what the private cloud really means …
As a thumb rule, the term private cloud means your service provider puts virtualization, automation, storage consolidation and all such services in your data centre and calls it the cloud. It is in your premises, and software vendors will sell private cloud managing technologies so that the IT guys can run their company’s data center.

And about public clouds…
This is the simplest cloud model. It a range of IDC services where a customer can create hybrid solution that may leverage collocation services, managed services and synaptic capabilities hosting and utility services hosting services. A public cloud, contrary to common misconception, does not mean all your applications and data is available for everyone to use or peek into. It is more apt to call it public service provider cloud. That’s when IT has gotten rid of their enterprise data center and migrated it all to the service provider’s cloud data centre.

Of public, private, clouds and vapours
The hybrid cloud or the virtual private cloud, as it is also called, means that IT utilizes both, the enterprise data centre as well as a public service resource. It is infrastructure versus services – I explain this with the Vapor vs. Cloud phenomenon; wherein Vapors denotes virtualization automated provisioning resources and standardization versus Clouds which denotes Common Location-independent Online Utility on Demand Service - the allegory is the same as having your own car versus renting one out.

On dispersed architecture or the Cloud Burst model…
People often think that the Internet and the cloud can somewhat be used interchangeably. Once you say cloud, it automatically means using the Internet to access the service. But in a scenario when enterprise data centers need to link to the cloud data centers, the Internet is not appropriate. The internet is best for ubiquitous access from a variety of devices – mobiles, laptops, smartphones, tablets, netbook. But for mission critical tasks, a dispersed infrastructure really requires more serious networking.

The dispersed cloud architecture
The dispersed cloud architecture revolves around several IDCs hosted at different locations. This becomes important because applications are becoming intensely interactive. The Web model has grown up, but technology wasn’t designed to facilitate this high degree of interaction. Even the opening light weight page of a search provider will have something like 29 objects on it. Millisecond latencies start building and we have a problem. The best way of getting around this is dispersed architecture. This will create localization of capabilities demanded by compliance or latency issues for highly interactive apps. It seems to be a smart practice of scale doing global business locally and dispersed architecture allows you to do that.

Cloud Cloud everywhere, not an adopter in sight
As with any new technology, there is adoption life cycle with different uptakes of different services depending on the different market segment. The largest global enterprises are in the midst of the classic technology adoption cycle– the awareness period, followed by the education period, which is then followed by questioning and skepticism. And then there comes news of success- some early adopter has dramatically reduced costs.