Yes, the 5G User Plane Can Be Cloud Native

As mobile operators contemplate 5G core network strategies, applying cloud native principles to 5G core functions is the best way to achieve network efficiency, flexibility and performance. Indeed, the industry seems to be in violent agreement that cloud native is the way to go. While this is a relatively easy case to make for 5G control plane functions, it’s a more difficult sell for the user plane. At Metaswitch, we think operators can have the benefits of cloud native in the user plane too and not have to rely on specialized hardware for core network switching and routing.

We developed a packet processing platform, called Composable Network Application Processor (CNAP), that overcomes the qualms about a software-based user plane and matches the performance of switching silicon for data plane functions deployed in containers on industry standard servers.

Cloud natively designed software applications hold many benefits for operators, such as operations automation, portability, dynamic scalability and n+k redundancy. These benefits are becoming better known as more operators get on board with this way of designing applications for the cloud, especially for 5G networks, and thanks to the work of groups like the Cloud Native Computing Foundation.

Applying cloud native principles to 5G control plane functions like the Access and Mobility Management Function (AMF) and Session Management Function (SMF) is well understood. Indeed, five years ago we built our cloud native virtual IP Multimedia Subsystem (vIMS) solution in this way.

But the 5G user plane is another story. The user plane handles all the traffic coming into the network from all end user devices and performs a lot of work on those packets to add service value – i.e., applying policies for video optimization, managing sessions, routing traffic to edge servers. The challenge is designing packet processing software that can be deployed in containers and scales elastically but also doesn’t consume so many CPU resources that it becomes too costly to run.

That calls for aggressive optimization to perform packet processing in one piece of software that touches each packet only once on its way through the network edge. And that’s what we’ve achieved with CNAP. It’s certainly a challenging problem, but it is solvable. And we have solved it. 

We think this development is really a breakthrough for the 5G core network. Admittedly, it is complex, so if you’re looking for a deeper dive into the topic, check out our recent articles and videos about the cloud native 5G user plane: