Software-defined networking has great promise, and continues to see strong investment. But it recently hit rock bottom on the Gartner Hype Cycle, and SDN to date has failed to meet its promise due to inadequate equipment, an incomplete standard, the lack of real-time control and of true industry support, and limited use cases.
That’s the assessment of Hayim Porat, CTO of ECI, who lays out his thinking about SDN in this Feb. 2 RCR Wireless article.
Porat suggests that standard appliances using application-specific integrated circuits for forwarding and perform as needed for the tasks at hand simply do not exist at this time. The lack of suitable solutions on this front for cloud and mobile environments, which Porat says are especially complex, is particularly acute, he suggests.
Although there is a wealth of standards groups out there, Porat says there’s much work to do – and competing interest to reconcile – to provide viable SDN solutions for this industry.
As for the lack of real-time control, Porat says that’s due to what he describes as “the industry’s naïve idea of a single, centralized controller.” He explains that there’s currently no solution “that can fully mediate between the application layer (switches, routers, firewalls and load balancers) and the physical network layer in a manner that is fast enough to enable dynamic networking in real time.” In fact, he says, using a centralized controller at this point would serve only to slow down traffic.
While just about everyone seems to have jumped aboard the SDN train, Porat notes that true interoperability on the software-defined networking front is far from reality. Compounding that problem is the fact that ROIs are muddy, which he says makes it difficult for carriers to abandon their proprietary implementations.
Speaking of implementations, he adds that Google and a handful of others are the only ones that have done complete implementations of SDN, and that all of those have focused on the data center and took years to complete.
All that said, Porat believes the adoption of SDN “seems inevitable” but at the same time advises that it’s time to rethink SDN by concentrating on creating a truly open network, enhancing the standards process, and revising the centralized controller concept.
Edited by
Kyle Piscioniere