In 2023, three years after this “big bang” event, cable operators continue to grapple with several challenges in supporting the surging traffic stemming from their commercial triumph. Factors such as network scalability, resource efficiency, network availability and operational efficiency are particularly critical at their IP aggregation layer.
Traditional IP aggregation architectures rely on monolithic chassis solutions, tailored to specific requirements at each geographic location considering 3-5 year growth projections, power constraints, space availability, and more.
When a chassis reaches its capacity limit, operators typically consider two options:
- Upgrading to a higher capacity chassis solution, commonly referred to as “forklifting”
- Implementing a hierarchical spine-leaf (Clos) architecture
Both strategies, however, introduce a range of technical, operational and economic challenges that cable MSOs strive to minimize.
In response to these challenges, the Open Compute Project (OCP) has proposed an alternative approach by defining the distributed disaggregated chassis (DDC) white box architecture. This architecture, which was developed by Tier-1 service providers, is supported by an extensive community of operators and vendors through the OCP. You can learn more about the latest (version 3) DDC specification here.
What is a typical cable aggregation network architecture?
A typical cable network architecture is evolving rapidly due to the increased demand for residential service aggregation. This growth is propelled by the expansion of services/solutions such as cable modem termination systems (CMTSs), virtual customer premises equipment (vCPE), and video streaming. Cable operators are focusing on minimizing impact between residential and business services and reducing north-south traffic for east-west traffic within primary hubs.
In parallel, the rise in Distributed Access Architecture (DAA) and deeper fiber deployments necessitate the introduction of super-aggregation layers. These layers are essential to mitigate the strain on the transport network. Advances in the blending together of optical and packet transport technologies (e.g. integrated ZR+ optics) allow for some flexibility in the overall size of the aggregation domain; however, the persistent increase in consumer broadband traffic necessitates highly dense aggregation solutions.
Moreover, services/solutions that were once centralized in high-scale data centers are now transitioning closer to the network edge to boost performance and efficiency. This shift is evident in network-based services/solutions such as virtualized edge tasks, content delivery networks (CDNs), and enhanced network services aimed at end users. As edge cloud services continue to proliferate and the demand for aggregation connectivity surges, cable MSOs are recognizing the pressing need for different and new scale-out network architectures.
What is the Distributed Disaggregated Chassis (DDC) model?
The Distributed Disaggregated Chassis (DDC) framework leverages well-known Clos principles that have been the foundation of the networking sector for the past forty years:
- Network processor-based distributed forwarding
- Element interconnects through a predictable fabric
- Expandable control-plane capabilities over general-purpose x86 CPUs
- Highly compatible, standardized optical connectivity
At a very high level, the DDC specification simply replicates any given chassis-based router on the market today, with one fundamental difference – DDC systems are composed of three discrete, interoperable units.
Each module fulfills one of three dedicated roles:
- Packet unit (“line card” equivalent): supporting user external connectivity through standard optics, with high-capacity packet-forwarding capabilities
- Fabric unit: used solely for interconnecting packet chassis in the system, the fabric in a DDC architecture is cell-based, deterministic, and input-queued
- Control unit: standard x86 server(s), running protocol operations and maintaining system functions
|Chassis Model||DDC Model|
|Proprietary line card||Standalone white box based on multiple chipsets and ODM vendors|
|Proprietary fabric card||Standalone white box based on multiple chipsets and ODM vendors|
|Proprietary backplane connectivity||Standard external cabling offered by multiple vendors|
|Proprietary power, fan, controllers common for all cards||Each white box comes with its own power and fan|
|Tens of SKUs on average||Six of SKUs on average|
Source: Based on information from Society of Cable Telecommunications Engineers (SCTE), a subsidiary of CableLabs.
In practice, DDC systems function and appear nearly identically to traditional, chassis-based systems once deployed. Their operational methodologies are closely aligned, as DDC systems can also function as a virtually singular entity managed by system orchestrators (e.g., DNOR). While the physical form might be different, DDC maintenance and troubleshooting are even simpler through component-based visibility and advanced automation.
How does DDC compare to traditional monolithic chassis-based architectures?
The notable distinction between DDC systems and traditional chassis architectures relates to vendors in the hardware and software domains. Typically, the vendor providing the DDC network operating system (e.g., DNOS) will differ from the hardware supplier responsible for the white boxes, optics, and cabling. This separation allows vendors to concentrate on their areas of expertise, innovate at a more organic pace and deliver at a more natural cadence, compared to vertically integrated systems, in which hardware and software must be developed together. It also enables operators to avoid vendor lock-in associated with traditional chassis systems.
DDC: the future of Cable MSO network infrastructure
The Distributed Disaggregated Chassis (DDC) architecture offers substantial benefits for a variety of network use cases. In this blog post, we’ve explored a few attributes of the DDC model and their alignment with the aggregation and scale-out requirements of cable MSOs.
Key takeaways include:
- DDC compares favorably, or even surpasses, chassis-based solutions in quantitative areas such as port density and power consumption.
- A white box software ecosystem exists not only for aggregation and core, but also for more complex solutions such as mobile aggregation, peering, and broadband network gateways or broadband remote access servers (BNGs/BRASs).
- There are significant technical (architectural) and business impacts thanks to DDC, including unprecedented flexibility in physical placement, power consumption planning and vendor selection.
Cable MSOs contemplating scale-out architectures ought to give DDC careful consideration. With its considerable and favorable influence on network design and architecture, DDC architecture incorporating white box solutions presents a powerful and attractive alternative for the future of network infrastructure.
Which Network Architecture Is Right for You