Network convergence has been discussed in the broader telecom industry for nearly three decades. But only in the past few years has it become a key issue for multiple systems operators (MSOs), which are tasked with supporting multiple services: business, mobile, and residential broadband.
These services are generally carried over separate networks, managed by distinct network monitoring systems and, most importantly, live in their own priority structure. They have their own service tier, service-level agreement (SLA), latency requirements, and so on. The hurdle is then to converge these services without disturbing the operational status quo that has allowed these services to flourish independently.
Recent developments in digitizing cable access gives greater opportunities for MSOs to move towards comprehensive network convergence. Network convergence will span the coexistence of multiple services over the same fiber, forwarding tables, data plane, server, etc. But getting to that point will take time and careful planning; even getting started is enough to make many network planners hesitant. To help support the shift towards full network convergence, MSOs need a robust framework.
Where to begin?
One possible approach to network convergence is to mimic the principles of 5G and, more specifically, network slicing.
5G was built to transmit and manage a collection of subservices in which each use case is likely to have a different need or functional expectation from the network’s available resources. For example, a use case like a robotic factory would have different needs than a self-driving car stopped at a traffic light, and different needs again than two family members in the same household where one might be watching a Netflix show and the other is on a Zoom call at the same time, on the same network.
The unique set of resource needs required by each service creates a blueprint that defines that service. In terms of network slicing, a convergence of services is a collection of blueprints managed over a shared infrastructure. Further, if you break up the overall end-to-end network into different sections (such as routing, processing, core functions, frequency distributions, customer premises functions), then each section paired with the resources that belong to a particular service would be considered a "slice." Network slicing is then the end-to-end management of these slices, which facilitate converged services, each of which has a unique blueprint.
Here, we’re assuming the existence of a bird’s eye view of the network, meaning the provider would be able to read out and even control the resources throughout, in order to create these unique services. 5G makes this possible by creating open interfaces to the data plane and control plane implementation of the system and to functional items within the cloud native infrastructure. The bird’s eye view is really another term for a robust domain controller that can adapt network elements as needed.
Three principles to drive network convergence
End-to-end network convergence for different lines of service is the end goal, but reaching this end goal has its challenges. And it’s here that a framework is needed. With the right framework, MSOs can enable multiple slices including residential (DOCSIS and video), business services (SLA and non-SLA), and mobile (macro and small-cell).
So how do we build a framework for convergence in cable? I suggest implementing three basic principles. The first is around service prioritization. This involves recognizing that each service vertical will have a unique prioritization that will have to be implemented. For example: residential, mobile, or enterprise services.
Second is that within services there will be a distinction of tenants and applications that will eventually need a granular treatment through the network. These would be subservices and other additional subservices. For example, video and DOCSIS in the residential vertical, where DOCSIS might also have various type of tenants such as a small cell or a home security system. This is a complex nested universe of unique services and subservices! It therefore demands robust tools for network slicing. Without them, in scale and over time, the bring up, provisioning, and troubleshooting of the overall converged system would be nearly impossible to maintain.
Third, the principles mentioned above are transitional over the lifecycle of the service. Automation mechanisms to adapt the network and ensure services can coexist are therefore necessary. This is where artificial intelligence comes in: the ability to recognize patterns and proactively propose solutions becomes invaluable. These artificial intelligence tools use the same principles as the tools developed for proactive network maintenance and advanced DOCSIS profile management. They would scan the current state of the network and adapt for performance according to certain guidelines.
Hard and soft slicing: The best of both worlds
In the framework for network slicing there are two general techniques: hard slicing and soft slicing. An MSO can use one or the other, or both, depending on necessity.
Hard slicing involves providing a fixed set of resources that are dedicated to a specific service. This could take the form of fixed bandwidth through a switched network, or dedicated radio frequency channels, or even a certain amount of compute space on a rack. Hard slicing offers guaranteed performance with limited flexibility.
Soft network slicing involves creating a policy mechanism where resources are more fluid and can be allocated as needed. The network is nimbler and paired to more involved automation and adaptive techniques. Soft slicing does offer a more efficient use of resources. It could, however, necessitate an overall quality of serve (QoS) policy involving services and subservices, which could be difficult to develop quickly in an environment that originates from separate, vertically integrated services.
Ideally, MSOs can utilize a mixture of hard and soft slicing as needed. In this form, vertical services can maintain their throughput expectations and simultaneously partition QoS to their specifications, while sharing broader resources.
Reaching the goal of network convergence
Network convergence and the ability to support services using the same shared network resources is an essential need for MSOs today. However, for MSOs to reach the end goal of network convergence, they need a robust framework.
One possible approach is to leverage key learnings from the framework set by 5G for network slicing, which is end-to-end and is dependent on open or programmable interfaces for both network and core resources. MSOs can apply some of these key principles of 5G to create their own, unique blueprint for network convergence. This end-to-end adaptive structure will enable MSOs to converge services and share resources, while being sensitive to the demands and policies of each service. SLAs are met, customers are happy, and MSOs get all the benefits of sharing resources.
Fernando Villarruel is senior consultant for regional sales at Ciena.