Cable MSOs are evolving to a distributed access architecture (DAA) that distributes the PHY or MAC/PHY function of the CCAP/CMTS to the fiber node, with digital packet-based transport replacing the analog transport that previously extended to the fiber node, as shown in Figure 1.
Figure 1: Distributed access architecture
This, together with fiber deep, which pushes the fiber node closer to the home, and DOCSIS 3.1/ Full Duplex DOCSIS 3.1, is enabling cable MSOs to scale access network capacities to 10 Gb/s. Furthermore, cable MSOs are building converged interconnect networks (CIN) as platforms for residential, business, and wholesale services, including 5G transport. This all requires a major upgrade of the transport infrastructure.
Table 1: Layer 3 vs. Layer 2
Layer 3 and Layer 2 both provide options for this packet transport. Layer 3 offers a more scalable solution with the ability to support a much larger number of nodes. Layer 3 can also protect against multiple failures, and its resiliency mechanisms are less dependent on network topology than Layer 2 protection mechanisms such as G.8032 Ethernet Ring Protection and Spanning Tree Protocol. Layer 2 technology is often perceived as simpler and more cost-effective. However, while the initial network design and configuration for Layer 2 may be simpler, once the IP network has been designed and configured, ongoing operations are greatly simplified by the IP control plane. And while Layer 2 has historically been lower in cost than Layer 3, the price delta between the two is quickly diminishing due to merchant silicon network processors and disaggregated routing.
Disaggregated routing offers many benefits over traditional routing, including reduced vendor lock-in, more choice, faster innovation, and cost-effective scaling, all of which lead to lower CapEx and OpEx, as discussed in a previous blog, Is Router Disaggregation Inevitable? But what special requirements are needed to address DAA?
Figure 2: DAA aggregation scenarios
While cable MSO networks and the terminology used to describe the different sites varies significantly, Figure 2 shows a generic network structure from the POP/data center to the optical node. First is the POP/data center, with functions that may include broadcast head-end, video-on-demand (VoD) content hosting, internet peering, and connectivity to the long-haul network. Next is the primary hub, where the CCAP core would typically be located in a remote PHY deployment. Then comes the secondary hub, which provides a conditioned aggregation point closer to the optical node. Pre-DAA, the CCAP or CMTS and edge QAM would typically have been located at either the primary or secondary hub. The remote hub provides an outside aggregation point closer to the optical node, either in a street cabinet or, closer to the optical node, a chamber or pole. Finally, the optical node itself is where the remote PHY or remote MAC/PHY devices are located and where the fiber network meets the coax network, as shown previously in Figure 1.
Four common scenarios for building the CIN are shown. Scenario A includes 100G+ aggregation at a primary hub and both 100G and 10G aggregation at secondary hubs. Scenario B runs 10G all the way from the primary hub to the fiber node with no intermediate aggregation. Scenario C moves the 10G aggregator to the remote hub. Scenario D provides a variant on scenario C with 100G aggregation at a secondary hub.
Table 2: Special requirements for DAA
Disaggregated routing can provide an ideal solution for packet aggregation in the primary, secondary, and remote hubs, as well as the POP/data center. In the primary hub, stacking may be required for both scaling and high availability, while the POP/data center is more likely to require fabric-based multi-unit scaling. Stacking may also be used for high availability at the secondary hub. Temperature hardening is required at the remote hubs, and in the case of chamber/pole deployment, a compact form factor that will fit in a weatherproof clamshell enclosure is also required.
For more information on this important topic, see the new Infinera white paper “The Case for Disaggregated Routing in 5G and DAA Transport Networks.”