Implementing Nephio
_{How CSPs can finally deliver customer value with intent-based networking}

One way to do that is to launch and adapt connectivity functions on demand. For example, a healthcare customer may want a remote surgery client with 100% reliable connectivity. A digital sports broadcaster may want the flexibility to adjust levels of service, based on the popularity of the game being shown.

5G disaggregated networks can do this. But actually, delivering this on a routine basis is hard, since these networks bring with them huge complexity.

The network automation complexity problem – Nephio to the rescue

The complexity of 5G networks is far too high for humans, so software has stepped in. Each function is wrapped in a layer of software that controls what it does. These are hosted in the cloud, where all can be coordinated.

Once everything is softwarized, in theory, networks can become ‘intent based’ – i.e. the user defines what they want to achieve (e.g. give the customer 10x bandwidth on match days) – and a piece of software that knows the state of the network works out how to best reconfigure it to achieve that outcome.

In reality, this has proven hard to implement. Network function software is complicated, and in a disaggregated network, parts are not always designed to speak the same language.

Generally, the solution has been containerization using Kubernetes, where each function is wrapped in a ‘container’. An orchestration platform manages them, ensuring each container’s inputs and outputs are translated into a consistent language across the network, and they all operate in the right sequence.

The problem is that Kubernetes was designed for enterprise networks, and has never quite worked for CSP networks, which have different needs from other softwarized networks, e.g. IP address assignment. The result is that emerging 5G networks with Kubernetes implementations are likely to struggle with high-level automation.

Nephio – an open-source Linux initiative – aims to solve this. It combines APIs to improve communication between Kubernetes containers, frameworks for the network automation of these containers, and an automation control plane that works with these frameworks and APIs to seamlessly coordinate the entire network. In other words, Nephio overcomes the biggest obstacle to intent-based network automation for CSP workloads. Nephio’s Kubernetes-enabled reconciliation ensures the network state is maintained, with respect to changes on cloud, or underlying infrastructure.

How to leverage Nephio to deliver intent-based network automation for CSPs

That is all good in theory. But the reality is that CSP networks are complicated and Nephio still needs complex implementation into the network architecture.

For all this to work, the different parts of the network, which may come from different providers, must be configured for Nephio. Even when these network parts are already softwarized and cloud-native, they are often designed for older, less automated ways of doing things, and there is still a need to upgrade them to work in an automated network environment.

Nephio utilizes Kubernetes as a uniform automation control plane and simplifies automation for cloud infrastructure, workload (network function) resources, and workload (network function) configuration. Nephio uses open, extensible Kubernetes Custom Resource Definition (CRD) models for consistent implementation across the cloud infrastructure and network functions.

And you need to make sure this ecosystem works end to end – for example, ensuring the service orchestrator is interacting with the Nephio platform to execute the service intent. Network functions are increasingly complex, with a wide range of dependencies and interdependencies, so care and expertise are needed at all stages to implement Nephio without introducing errors.

Security must also be considered. Nephio automates the deployment and management of network functions, which is valuable, but can create new opportunities for attack. It is important to implement appropriate security measures to protect Nephio from unauthorized access and malicious attacks. This is why Nephio has a security group dedicated to aspects of CSP workload security while implementing the automation.

Finally, the deployment of Nephio will likely need training and the skills to use and maintain it.

Nephio implementations are a job for both Network Equipment Providers (NEPs) – who CSPs will increasingly demand are Nephio compliant – and for the CSPs themselves at a network architecture level. Addressing these challenges requires technical expertise, robust processes, and a deep understanding of industry trends and customer needs.

Nephio represents a major leap forward for CSP network automation and the ability to cost-effectively deliver high value services. CSPs and their suppliers should look at Nephio very seriously, and give thought to how to implement it effectively.

Capgemini: the best way to implement Nephio for your business

Capgemini is the only Systems Integrator in the initial group of Nephio members. We have a detailed understanding of Nephio, and CSP networks generally, and have developed frameworks for implementing Nephio and achieving Nephio compliance. We can partner with CSPs and NEPs to integrate Nephio effectively and securely.

By combining our years of experience in developing network function frameworks with a deep understanding of Nephio, we intend to help reduce the complexity of managing network functions –  bringing in the much needed Kubernetes-based automation push for CSPs.