Virtualizing Broadband Networks: Q&A with Tom Cloonan and David Grubb

The future of broadband networks is fast, pervasive, reliable, and increasingly, virtual.

Dell’Oro predicts that virtual CMTS/CCAP revenue will grow from $90 million in 2019 to $418 million worldwide in 2024. While network virtualization is still in its earliest stages of deployment, many operators have begun building their strategy for virtualizing one or more components of their broadband networks.

We caught up with CommScope’s Broadband Networks executives Tom Cloonan (CTO) and David Grubb (SVP of Engineering) to bring us up to speed on virtualization: What it is? Why is it so important for operators? And what’s driving their decision-making?

CLICK TO TWEET: In this blog, CommScope’s experts explains what network virtualization is and why it’s so important for operators.

What is network virtualization?

In its simplest form, network virtualization is the process of creating software functions to replace hardware functions that once ran in physical components within the network. For example, virtualized software can replace management plane functionality, control plane functionality, and data plane functionality that originally resided in the physical network.

The virtualized functions can then be relocated to operator private cloud environments in headends or data centers—and in some cases public cloud environments, such as Amazon Web Services (AWS), Google Cloud, and Microsoft Azure.

In general, the virtualized software is re-targeted for operation on Commercial-Off-The-Shelf (COTS) x86 server platforms. It could also be isolated and operated within a Virtual Machine (VM) environment and/or within a container-based environment. 

What are the benefits of network virtualization?

Virtualizing network functionality can offer many benefits:

1. Horizontal scaling: It permits operators to use horizontal scaling techniques to expand their network capacity, adding new servers whenever demands require it.
2. Vertical scaling: It permits operators to capitalize on the vertical scaling of processors, taking advantage of the advances in performance of processors within server platforms.
3. Modular architecture: Virtualization and Open Source code are both gaining traction in the networking software community, so many virtualized software solutions will use some Open Source code and open APIs—making it easier (and faster) for operators to add their own features. The use of published APIs also permits operators to employ traditional IT professionals to work on these network-oriented functions.
4. Increased innovation velocity: Modular design combined with a Continuous Integration & Delivery (CI/CD) enables new features to be developed, tested, and field-trialed quickly. If the new feature proves to be successful, it can then be rolled out across a deployment footprint rapidly.
5. Analytics: Virtualization opens the door to collecting more network data and performing more Analytics work on that data, leading to improved Quality of Experience (QoE) for subscribers.
6. Automation: Virtualization (when coupled with Artificial Intelligence software) also opens the door to full headend automation in the future.
7. Cost reductions: The use of COTS server platforms with a virtualized environment may also permit cost reductions over time.
8. Processing elasticity: Since virtualized software runs on general purpose processors, that same processing power can be re-used for different applications at different points in time to capitalize on variations in networking load throughout the day. 

What are the stages of transition to virtualized networks?

Each operator will choose to transition at a different pace. And each operator will also choose different elements to virtualize.

Virtualization can be used for management, control, and data plane, and it can also be applied independently to different applications in the network (e.g. video vs. data). Operators can choose one, two, or all of those areas to virtualize. For example, virtualization can be used for elements of DOCSIS, video transport, PON, Ethernet, Wi-Fi, LTE, 5G, or CBRS.

Any last-hop technology can have some of its functionality virtualized. But most operators will likely align with a chosen technology in their transition towards virtualization. In network applications, virtualization is often paired with disaggregation to create a more modular architecture—an example of this would be CCAP moving to DAA.

Virtualization of the access network requires operators to plan out their new network components and data paths. Functions that were formerly encapsulated in backplanes are replaced with an IP switching layer referred to as the Converged Interconnection Network (CIN). Some functionality will likely need to remain as physical network elements, and operators must intentionally select which functionality to keep physical and which to move to a virtualized option.

Next, traffic engineering work is key to creating a suitable end-to-end solution that provides high QoE for all subscribers. There is an important relationship between traffic engineering and the architecture of virtualized solutions. We recommend that operators select a strong vendor partner with deep traffic engineering expertise to work with as they plan these changes. 

Once operators architect their new network, lab trials are essential. Only after successful lab trials should operators begin their first field trials. After those field trials are successful, then full-scale deployments can begin. 

How do the stages of transition differ for operators of different sizes?

Today’s operators implementing virtualization are following many different paths.

Larger operators tend to have larger software teams that can provide some of the virtualized software functions—such as orchestration functions that launch and monitor the health and sanity of other software containers. Smaller operators tend to rely more heavily on the vendor community for all of their virtualized software sub-systems.

In either case, it is beneficial to operators to work with a single vendor for much of the solution or with a group of vendors who have proven interoperability between their virtualized solutions.

What components of virtualization should operators implement first (highest ROI/priority)?

This is a unique decision to be made by each operator individually. However, many operators are looking at following one of two paths for DOCSIS Distributed Access Architecture (DAA) solutions right now.

Some are looking to follow the Virtualized Core & Remote PHY Device architecture, which tends to virtualize management and control and data plane functions (for the DOCSIS MAC). Others are looking the follow the Flexible MAC Architecture (FMA) approach, whereby the management and control plane functions will be virtualized in a MAC Manager.

Both approaches heavily utilize virtualized software and offer tangible advantages for operators.

For video, no matter which of the two data paths is selected, changes will be required since the edge QAM function moves from the headend to the fiber node in both architectures. Selecting an approach for video virtualization that maximizes flexibility is an important aspect of the overall plan. 

How do CommScope’s virtualization solutions compare to the competition’s?

CommScope has a large number of virtualized platforms within its product portfolio. Some are already available and some are in development. This large array of interoperable virtualized platforms within a truly end-to-end networking portfolio is one of the most compelling reasons to choose our solutions.

Examples of our virtualized video applications in use today in multiple operator networks are: Vertasent for Switched Digital Video, MDC for IP Video Ad Insertion, and VUE for MPEG Transport Stream processing.

In the DOCSIS world, for example, CommScope’s virtualized solution includes the vManager and the vCore platforms, which connect over the CommScope ICX switches within the Converged Interconnect Network to CommScope’s Remote PHY Devices within CommScope nodes. This end-to-end solution provides guaranteed performance levels that have been well-tested and field-hardened in real-world deployments for over 20 years.

Much of this hardened software within the CommScope virtualized solution has been re-factored for use within COTS servers. The solution also utilizes micro-services techniques, making it quicker and easier to maintain and upgrade code. The resulting platform will undoubtedly yield industry-leading density and scale within the COTS server platforms of the future. CommScope is also developing strong resiliency capabilities within its designs, capitalizing on its many innovations that brought carrier-class performance into the DOCSIS world.

Our field-hardened code has ensured that the virtualized solutions we have in trials with some of the world’s largest operators have been incredibly stable in operation, and we’re adding features and scale continuously. Many of these will go into deployments in the second half of this year and in 2021.

For more information, please visit our virtualization portfolio. 

Additional resources

• Virtual Headend
• Video
• 10G Infographic
• Virtualizing the cable headend whitepaper