Bullpen: DIBA Reconsidered

The author of the well-written video over DOCSIS article in the November issue of CT discussed the potential advantages of the DOCSIS over IPTV bypass architecture (DIBA). The architecture and technologies proposed are certainly physically realizable.

However, before dedicating any resources to further refining and standardizing another infrastructure tied to the HFC network, the industry would benefit from a careful exploration of its pros and cons.

What follows is a summary of some of the potential disadvantages and related misconceptions of this proposal.

1. The basic premise behind the DIBA proposal is that cable modem termination system (CMTS) media access control (MAC) cores are currently expensive and will continue to be so in the future. This will soon be proven to be incorrect. Both the DOCSIS 3.0 and modular CMTS (M-CMTS) specification work identified cost reduction as one of their key goals. Many of the features added to the CableLabs specifications were aimed at driving cost out of DOCSIS. In addition, the arrival of multi-core processors, chip fabrication enhancements and other new technologies is also driving cost out of the DOCSIS MAC cores.

2. The article’s statement that "DIBA accomplishes the goals of the M-CMTS without the unnecessary expense of the M-CMTS" is somewhat misleading. In fact, DIBA is very dependent on the existence and availability of a normal CMTS somewhere in the vicinity of the DIBA edge QAM modulators. Why? The article states that a normal CMTS is needed to manage the operations of all of the DOCSIS upstream channels in the system. But upstream management is actually the most challenging and costly of all of the tasks within a CMTS.

The article also indicates that a normal CMTS is needed (1) to provide all of the primary downstream channels (with DTI synchronization) for all of the cable modems receiving from DIBA-based edge QAMs; (2) to manage all of the MAC management messages generated to and from every modem receiving from DIBA-based edge QAMs; and (3) to interact with the PacketCable Multimedia (PCMM) policy servers and with the edge resource managers (ERMs) while setting up L2TP tunnels between the video servers and the edge QAMs to support the establishment of IPTV sessions. DIBA also requires that a normal CMTS be available to manage dynamic bonding changes (DBCs), downstream service identifiers (DSIDs), and MAC domain descriptors (MDDs) to each of the modems receiving from DIBA-based edge QAMs. As a result, it seems incorrect to imply that DIBA eliminates the need for a CMTS.

3. DIBA proposes keeping only a small subset of the DOCSIS downstream MAC functions that are required to capitalize on well-behaved, walled-garden IPTV traffic and moving these remaining DOCSIS downstream MAC functions into the edge QAM. The first of these changes can be done in an IPTV-oriented CMTS MAC core as easily as it can be done in an IPTV-oriented edge QAM. (This was termed "DOCSIS-LITE" in an ARRIS proposal from several years ago). The second change amounts to a simple shell game, moving the minimized DOCSIS MAC functions (and costs) from one box to another. The cost of implementing those DOCSIS downstream MAC functions will be incurred whether the function exists in the MAC core or in the edge QAM. The DIBA proposal cannot claim that the cost goes away if the functions are simply moved into the edge QAM.

4. The article argues that the use of DIBA can help to minimize system costs by minimizing the amount of "hair-pinned" DOCSIS downstream external-PHY interface (DEPI) traffic that flows from the M-CMTS MAC core to the edge QAM over the converged interconnect network (CIN). If that is truly a concern for any cable operator, then that operator can always choose to deploy an IPTV-oriented integrated CMTS (I-CMTS) implementation (which entirely eliminates DEPI traffic) instead of an M-CMTS implementation. However, the dropping costs of Ethernet links and Ethernet switches will probably make worries about CIN costs a moot point in future M-CMTS systems anyway.

5. The industry will have to expend substantial energy and resources on defining a new CableLabs DIBA standard if it is selected as the primary technique for delivering IPTV. That will consume time and effort – unnecessarily so, if DOCSIS already can provide the desired IPTV delivery functions with the lower costs that are required for the future. Again, developing a new standard would take time.

6. For years, cable operators have been trying to minimize complexities and redundancies in their back-office networks to reduce operational expenses. The DIBA proposal bucks that trend by introducing yet another architecture for delivering data into the HFC network. In addition to DOCSIS, PacketCable/PCMM, and MPEG-TS delivery systems, DIBA would represent a fourth headend infrastructure that would need to be specified, managed and maintained.

7. The impact on DOCSIS provisioning equipment and configuration scripts is unknown and could be extensive.

8. It is questionable if legacy edge QAMs would have adequate processing power in their available circuitry to add in the requisite DIBA functions. DIBA does port a subset of DOCSIS functions to the edge QAM, and some processing hardware must be added or made available to implement those DOCSIS functions.

9. By eliminating many of the DOCSIS quality of service (QoS) functions, DIBA edge QAMs will be limited to supporting only IPTV feeds from trusted sources within the cable operator’s back office. As a result, DIBA edge QAMs will likely not be able to support IPTV feeds from third-party content providers with whom cable operators might have arranged business deals because these feeds will likely have to be treated as coming from an untrusted source, requiring the application of QoS that DOCSIS offers. As a result, cable operators might be better off using standard CMTSs that can implement DOCSIS-LITE functions (without QoS) for IPTV feeds from trusted sources and full DOCSIS functions (with QoS) for IPTV feeds from untrusted sources. This arrangement gives cable operators more flexibility in choosing the sources of their IPTV content, while still utilizing all available DOCSIS headend equipment.

10. Studies indicate that the burstiness of IPTV traffic can lead to great stat-mux gains if it is delivered over wider channels. As a result, there are great efficiency gains that can be realized if IPTV is transmitted over DOCSIS 3.0 bonding groups. If the complexities of channel bonding are ported to the DIBA edge QAM, then the added functionality will likely add substantial costs to the edge QAM. As a result, the original goal of DIBA (to reduce cost by eliminating DOCSIS functions) is effectively erased if cable operators decide to capitalize on the IPTV stat-mux gains offered by DOCSIS 3.0 channel bonding.

While DIBA offers potential benefits, the preceeding list illustrates that it also introduces many disadvantages. In the opinion of the author, an IPTV delivery system based on standard DOCSIS, especially if hooks are added to accommodate DOCSIS-LITE functionality within the CMTS for trusted IPTV content, offers more advantages.

Tom Cloonan is chief strategy officer of ARRIS. Reach him at [email protected].