SDN: The Key to Computing in the Cloud
Interop this year seems abuzz with talk of the cloud-based data center. But whether you call this Datacenter-as-a-Service (DaaS), data center virtualization, the software defined datacenter or any other title, the fact remains that software defined networking is what makes it all possible.
Earlier this week on my blog at ENP's sister publication IT Business Edge, I highlighted the possible rise of the mega datacenter (megaD in the latest parlance). The idea is that, before long, most enterprises will shed their on-premises infrastructure in order to consume data resources utility-style, with massive regional data facilities springing up like today's power plants. The precedent for this is, in fact, the power industry itself, which converted from factory-based generation to on-demand utility delivery nearly 100 years ago.
Data is not electricity, however. It is only with the advent of SDN that the idea of a fully scalable, end-to-end, software-based infrastructure can be realized. And now that that flexible networking technology is here, the race is afoot to deliver on its promise for cloud-based compute environments. Google, for example, recently added a new SDN component, dubbed Andromeda, to its Compute Engine platform, which the company says will allow complete computing environments to tap the immense scalability of Google's massive data infrastructure. Ultimately, the entire service will become transparent to the user: simply request more resources, and server/storage/networking capabilities will rise to meet the challenge.
But just because top providers like Google are ready to delivery this kind of functionality, does that mean the enterprise is ready to buy? According to Ziff Brothers' Eric Lundquist, the answer is yes. When it comes to lower capital and operating costs, automated management and the ability to meet the demands of Big Data, M2M communications, mobile traffic and all the other forces hitting the enterprise, software-based infrastructure has arrived just in time. At the same time, logical constructs are more suited for next-generation data applications like collaboration and on-demand services.
SDN alone, however, will not be enough to produce the digital nirvana that many organizations are striving toward, says Cloud Computing News' Geoff Bennett. There is also bandwidth and automation to consider, particularly once we start talking about widely distributed networks of interconnected data centers. With cloud deployments increasing 30 to 50 percent per year, the ability to define "super channels" within available bandwidth, coupled with the automated transport layer that SDN provides, will be crucial to accommodate the rapid service provisioning that knowledge users will come to expect.
Clearly, the idea that "the network is the computer" is giving way to "the network is the data center." In an era when entire computing environments can be spun up on a whim, functionality will derive mainly from the ability to harness disparate resources into a cohesive whole.
And for that, you'll need a network architecture that is scalable, flexible and comes in under budget.
Photo courtesy of Shutterstock.