Title: Principal Architect, DOCSIS Technology, Cox Communications
Broadband Background: Bekele is responsible for the evaluation and deployment of the latest DOCSIS technology within Cox Communications. He is currently architecting Cox’s migration to DOCSIS 3.0/M-CMTS, working on the intelligent management of DOCSIS resources to increase network efficiency, and modeling of the DOCSIS network to accommodate new and advanced DOCSIS services and analysis of cost impact of new services. And if all of that’s not enough, Bekele was a panelist at the recent SCTE Conference on Emerging Technologies and a panelist on a recent Webcast on DOCSIS 3.0 presented by Cisco and CT. (Registration is required to view the Webcast.)
What do you see as the timeframe for DOCSIS 3.0? Do you think we’ll see any new CMTS or silicon vendors?
We expect to have fully compliant DOCSIS 3.0 modems by the first half of 2008 and fully DOCSIS 3.0 compliant CMTS by the second half of 2008.
We will see new CMTS vendors for DOCSIS 3.0. M-CMTS has also opened the way for new vendors to participate on the downstream RF side. And the introduction of the modular upstream PHY interface in the future should allow more vendors to participate on the M-CMTS core side. I believe once we resolve the resource management piece on QAM sharing between DOCSIS, SDV and VOD, we will also drive the existing vendors to modernize the CMTS core chassis.
In a recent CT Webcast you said IPv6 wasn’t as big a priority for Cox as other MSOs because of how the network is architected. Can you explain that?
One of the driving forces for IPv6 is the scarcity of private IP addresses. In Cox’s case, all our private addresses are local to each regional distribution center (RDC), so we will not run out of IP addresses anytime soon. We have chosen a distributed architecture for both the provisioning and management tools. Our tools are architected to be able to get to all the devices from our central NOC in Atlanta. Unless we see a service that requires IPv6, we will be working on the parts of DOCSIS 3.0 that will provide more speed and enable advanced services such as multicast QoS.
Why is Cox a proponent of M-CMTS?
M-CMTS gives us a natural growth path that allows us to leverage our existing CMTS architecture and grow our downstream capacity. The current 1×4 MAC domain configuration has left a lot of upstream channels stranded. M-CMTS provides that downstream growth at a substantially lower cost per downstream port. As we increase speeds and need more capacity, instead of adding new line cards, we just add a U-EQAM (Universal EQAM). This also allows us to build an architecture that will enable DOCSIS to share QAM resources with switched digital video (SDV) and VOD, which will reduce the QAM dollars spent per service as well as the capital spent per bandwidth ($/Hz) in the future. In the long run, QAM sharing allows us to separate resources from applications/services and go to a bandwidth/resource on demand architecture where services can request bandwidth on an as-needed basis.
You were a panelist at ET in Houston; what can you tell us about Cox’s efforts in the realm of service velocity?
We are working across data, voice, and video groups to see how we can leverage existing infrastructure to generate more revenue and get to a converged architecture that allows us to seamlessly add new services.
For example, for L2VPN services, we are working with CableLabs, other MSOs, vendors and other standards bodies such as MEF to come up with ways to automate the provisioning process and easily manage the service. The goal is to create a modular framework so we would not have to implement custom management systems and provisioning tools for every service delivered. We are pushing hard to achieve that with L2VPN commercial services today.
How can a cable operator take the first steps to implementing Intelligent Resource Management, and why is it important?
There are several aspects of Intelligent Resource Management (IRM). The first one is the ability to collect accurate data from the network in a standardized, more efficient way, and IPDR provides that in the DOCSIS world. The ability to abstract the application from the resources and be able to go to a resource on demand (RoD) model is also critical to doing IRM. The policy server in PCMM gave us the ability to separate applications from DOCSIS resources by allowing applications such as VoIP and gaming to dynamically request DOCSIS resources, creating an application on demand (AoD) architecture. ERMI in M-CMTS will provide Universal Edge Resource Management (UERM) to create the separation of services and resources at the edge and allows SDV, VOD and DOCSIS to request QAMs dynamically. Once you have the above pieces in place, you would need to bring in management tools and algorithms to optimize the network efficiency, improve the customer experience and leverage that same network to generate more revenue.
IRM is important because we are moving from service silos to an architecture where services are sharing the same resources to deliver to the same pool of subscribers. For example, the introduction of PacketCable VoIP was our first foray into convergence for cable. This required us to keep track of peak call periods, simultaneous calls and set admission control policies for data and VoIP accordingly. PacketCable Multimedia applications are now being implemented, and these will require real-time knowledge of the network status before making admission decisions.