DENVER — Some top cable executives insist that virtualizing the cable network is not more complicated than virtualizing a telecom network. But as a reporter attempting to cover network virtualization, I disagree.
“There’s not much difference between us and the telcos,” said John Dickinson, VP of advanced commercial engineering at Charter Communications, speaking at a Light Reading breakfast entitled “Virtualizing the Cable Architecture” at this week’s SCTE-ISBE Cable Tec Expo.
And in a conversation recently with Noam Raffaelli, Comcast SVP of network and communications engineering, he said, “We know what the telcos are doing. There’s not that many differences. In service virtualization we want to get to the same place, to decouple the underlying network from the overlay services. The underlying differences are in the access.”
Therein lies the rub. The access.
Cable’s outside plant includes lots of hybrid fiber coaxial cables (coax), which telcos don’t use. And the delivery of video as well as Internet over the cable plant involves complicated engineering. You start having to deal with acronyms like Data Over Cable Service Interface Specification (DOCSIS).
In addition, cable operators are in the process of replacing their old cable modem termination systems (CMTS) with the new converged cable access platform (CCAP).
At this week’s Cable Tec Expo there was lots of interest in two other difficult-to-grasp acronyms — the new remote PHY and remote MAC/PHY technologies. A session on Tuesday about R-PHY was so popular that event organizers had to open two more rooms to accommodate the crowd of more than 400 people.
R-PHY and R-MAC/PHY will allow cable operators to move functions out of the “big iron” boxes in their cable headends and put them in smaller devices out in the cable access network, closer to end users. In terms of virtualization, these technologies help cable operators to decompose aspects of their networks, pushing physical parts out into the cable plant with the end goal of running it all with software.
Some vendors, such as Casa Systems, are referring to their R-PHY technology as a virtual CCAP. And Nokia is bragging that its R-MAC/PHY can replace the CCAP box in the headend, entirely.
But wait, isn’t CCAP, itself, a new technology that all cable operators are transitioning to? Why is there talk about virtual CCAP and dispensing with CCAP?
CCAP vs. Remote PHY
I asked one top cable executive at the show this week if there was a conflict between CCAP and the R-MAC/PHY trend, and he said framing the question in that way was frowned upon within the cable industry. This executive, who asked not to be named, said the transition from monolithic devices in the headend to a distributed, virtualized architecture will follow a progression that could take a decade.
“CCAP is a short-term money maker,” he said. “Short term in cable terminology is 10 years. CCAP lets us make an investment today and realize revenue for a long time. Because of its inherent modular nature we can break pieces out and start virtualizing pieces.”
This argument makes sense: cable will gradually virtualize its access architecture.
Simultaneously, some cable operators, such as Comcast, are working with the Central Office Re-architected as a Data Center (CORD) projects. These open source projects aim to make telco central offices and cable headends look more like data centers, which can run software-defined networking (SDN) and network functions virtualization (NFV) in standardized ways.
CORD software takes traffic from disparate technologies and converts it into Ethernet packets. Then, the packet switching that needs to be done at the central office or headend can be moved onto commodity hardware and controlled by one operating system.
At the Cable Tec Expo breakfast this week, Charter’s Dickinson said, “Being a fast follower has been beneficial to the cable industry. It’s becoming more of a timing issue for us to engage and how to deploy this.”
Still, circling back to my original point, it seems as if the cable industry will have to deal with a lot more complexity to virtualize its networks because of its access: wrangling with DOCSIS, R-PHY, R-MAC/PHY, and CCAP.