Pipeline Profile: Jeremy Bennington

Title: Business Development Manager, Symmetricom

Broadband Background: Bennington is responsible for business development of synchronization technologies and markets for Symmetricom. His work includes development of the DOCSIS Timing Interface (DTI) and several other synchronization technologies for the broader communication market including IEEE1588, Synchronous Ethernet and the Universal Timing Interface. He has also served as a contributing author of CableLabs’ M-CMTS DOCSIS Timing Interface, a member of Cablelabs’ DOCSIS 3.0 specification team, a contributing author of CableLabs’ Commercial Services Specification and a member of the ITU SG15 Synchronization Standards Group. Prior to Symmetricom, he held various positions at Intel, Tele-Communications Inc. (TCI) and Purdue University. He holds a Bachelor’s in Computer Engineering and a Master’s of Science in Management from Purdue University.

Your recent CT article deals with synchronization, a topic that hasn’t received a great deal of attention in the past. Why is that?

Synchronization in the past was limited to the SONET transport used by some cable operators between headends, or NTP used for management applications or PacketCable. Today these applications still need synchronization, but the introduction of the more efficient and flexible M-CMTS requires a new synchronization interface called the DOCSIS Timing Interface (DTI). With this new requirement, business services (T-1/E-1) and the growing need for NTP for IP applications, synchronization is becoming a more important topic.

The article makes it sound like T-1 and E-1 emulation require more stringent timing than DOCSIS does. Could you elaborate?

The requirements are a little different. T-1/E-1 synchronization requires frequency stability with respect to a global reference; in the telecom market they call it Stratum 1 traceability. DOCSIS M-CMTS requires time and frequency traceability to a local reference (a Root DTI Server). The M-CMTS requirements are more stringent than T-1/E-1 within a headend to ensure that edge QAM and CMTS equipment appear to the cable modem as if they were one shelf. The global requirements for M-CMTS and T-1/E-1 are very similar.

How do multiple synchronization tiers help achieve economies of scale and conserve capacity and throughput?

The multiple tiers of synchronization help a cable operator choose what synchronization architecture to deploy. For an operator who does not need to support telecom services such as T-1/E-1, does not have a large network and does not want to measure QoS, Tier 2 synchronization is the most cost-effective choice. Tier 1 synchronization enables an operator to support all telecom requirements including T-1/E-1, but also other potential applications like mobility. It also ensures that QoS between sites can be measured and enables further scalability of the M-CMTS. Tier 1 synchronization ensures the best scalability and ensures efficient use of capacity and throughput. Tier 3 synchronization is limited to the old integrated CMTS architecture. Tier 3 is less economical, scalable and cannot support the same throughputs as Tier 2 and Tier 1.

It sounds like timing could be an Achilles’ heel of the M-CMTS architecture. Is that right? If so, what’s the solution?

Synchronization is certainly a vital part of the M-CMTS architecture, but there are several other pieces that are just as important. Synchronization must be reliable and precise; without it the M-CMTS cannot operate. To ensure that synchronization is not lost, the DTI is by design extremely robust. DTI servers can be deployed with internal redundancy where the critical guts of the server are backed up internally, so if they fail, the outputs of the server are not interrupted. Multiple DTI servers can be deployed to offer redundancy as well. On the DTI client side, most edge QAM and M-CMTS equipment manufacturers have implemented a primary and secondary DTI port to allow for path protection. Moreover, DTI is continuously measuring, monitoring and reporting its health so the cable operator knows at every moment how things are going. So there are a variety of ways to deploy DTI to ensure it does not become the Achilles’ heel of the solution and not cost an arm and a leg.

There always seems to be some debate on the merits of centralized vs. distributed architectures: VOD, powering and now CMTS. What’s your take?

We could extend that debate to government, education and several others, but for now I’ll talk about DOCSIS and VOD. When we talk about DOCSIS and VOD, we are really talking about the cable operator’s primarily mechanism to deliver IP services to the subscriber. Since IP applications are growing at an infinite pace, it is critical for operators to ensure they can retain the subscriber and offer new services without increasing their cost at the same rate. Scalable architectures like M-CMTS are the only way to ensure the cable operator can grow economically and flexibly.

Could you give us some background on Symmetricom’s involvement in the development of the DOCSIS Timing Interface (DTI)?

Symmetricom has been involved with M-CMTS and DOCSIS 3.0 since the beginning. We were one of several vendors to propose a solution for the DOCSIS Timing Interface and were chosen by CableLabs to be the primary author of the specification. Symmetricom is the market share leader in synchronization technologies globally. We developed the fundamental technology that DTI uses many years ago for a proprietary application. Symmetricom’s primary goal in all markets is to sell a synchronization server like a DTI server, NTP server or BITS. When the cable industry came to us with the need for DTI, we decided to grant a royalty free license of our intellectual property through CableLabs to allow the industry to use this extremely precise and cost effective solution. Since then, DTI has been specified, and at least 10 manufactures are integrating DTI into their products. Symmetricom hopes that through our leadership in this area we can help the industry offer new services, economic architectures and win the battle for the subscriber.

Is this more of a topic of concern at corporate engineering? Or has it moved to cable operators’ divisions or regions? Do headend engineers and technicians need to understand this now?

Right now, the topic has a foot in the corporate office and in the regions. Since this is a new opportunity to offer enhanced services like DOCSIS 3.0 bonding and do it economically, everyone is interested. Most operators will make a corporate decision to go with M-CMTS and then leave it to the regions to decide where and when. Some operators are different, but both corporate and headend engineers need to understand the fundamentals of M-CMTS, including DTI and the IP aspects of the architecture. Like anything new, it is an opportunity for engineers to learn, tinker and become valuable experts.