The goal of 5G deployments is to provide consumers and business users with highly engaging experiences that leverage the high reliability and high performance provided by the underlying network. The challenges from a technical perspective then become twofold – support a diverse set of use cases without adding to operational overhead. And once the network is setup, as a collection of cloud-native services (or CNFs (Container Native Functions) that communicate over well-defined interfaces, the operator needs to support a range of Day 2 and Day 2+ operations such as node provisioning, scaling, updates, monitoring etc.
One of the previous blogs discussed what a real-world 5G Radio and 5G Core platform deployment looks like – https://www.vamsitalkstech.com/5g/real-world-5g-deployment-architecture/.
5G mobile broadband network operators transitioning from 4G have scale challenges to deal with. The 5G mobile network will be standards-based and controlled by software. Further, they will need to distribute compute workloads that manage (tens of) thousands of infrastructure elements across hundreds of (edge) data centers and a few central data centers. The compute elements will range from radio networks to those running core functions. All of these services will need to operate under strict SLAs and requirements for latency, uptime and performance.
The typical deployment model of 5G networks is shown above (courtesy AWS). Radio sites running 5G, a fronthaul/midhaul/backhaul network, a core network site, and a telecom/IT data center. A virtual Network Operation Center (NOC) can run in the cloud and that hosts the main control plane as well as Operations Support System/Business Support System (OSS/BSS).
The goal in moving 5G functions to software is to not only ensure that these requirements can be met but also to ensure seamless management of these functions across various layers – edge, access, and core data networks. Most of these network functions will run on containers as CNFs (Container Native Functions) along with some VNFs (VM-based functions).
The edge or MEC (Multiaccess Edge Compute) hosts network functions such as virtualized/containerized CUs (Central Units), functions such as DNS, user plane functions (UPFs), local domain name systems (DNSs), and some local application functions. Where far edge data centers are concerned, they host virtualized or containerized DUs and RIUs. Each of the above locations can also contain physical infrastructures such as routers and commodity servers that host NF virtual infrastructure (NFVI) pods. These nodes are provisioned to be able to host virtual network functions VNFs as well as CNFs. A mobile backhaul connects all backhaul traffic that runs between the edge sites and the central data center(s).
The Requirement for Network Automation
5G is characterized by a few defining features and requirements which call for pervasive automation:
- Deploy services across new sites seamlessly by enabling verification of new CNF versions, enabling easy change of configurations.
- Support design for resiliency both from a network and human perspective
- Support Day 0 automation with the core network orchestrator- Provisioning and lifecycle management of all VNFs/NFs in the mobile network. As applicable support CI/CD pipelines that achieve the same
- Support Day 1 automation with configuration management of all VNFs/NFs across hundreds of sites – in a phased manner. Typically accomplished using IaC (Infrastructure as Code) tools such as Terraform, Ansible etc. This also may include configuration of mobile backhaul elements as necessary /
- From a data center perspective, support the configuration of data center fabric to enable the provisioning and successful communication of all VNFs/NFs
- Managing and onboarding new virtual and physical network elements onto the network, as and when they are introduced as covered here- https://www.vamsitalkstech.com/5g/so-what-are-the-key-management-capabilities-for-mec-edge-platforms/
- Finally Day 2+ automation includes vendor and systems upgrades of the nodes running VNFs/CNFs, incremental configuration automation, NF monitoring, scaleup/scale down based on usage and service assurance
Conclusion
Infrastructure elements ranging from hundreds to millions, varying by the size of the deployment, these include both radio functions as well as core. Typically these are servers running RAN workloads (DUs, CUs) or other UE (User Equipment). Deployments that span different locations. Ideally, operators want to enable lean operations teams that can roll out networks running hundreds to thousands of NFs quickly by deploying a large number of sites in as low a time as possible. Considering the complexity of the distributed architecture and the deployment, monitoring, and maintenance of all these NFs, it is inevitable for the operators to have a fully automated solution for network deployment, operation, and management with the following tasks for the automation network. One of the subsequent blogs will discuss GitOps as a technology approach for achieving this- https://www.vamsitalkstech.com/cloud/gitops-the-key-enabler-of-cloud-native-platforms/. We will delve into the architecture of a 5G platform deployed and updated using GitOps. Before we do that, the next blog will introduce the killer app in 5G industry use cases: Network Slicing.