For most production deployments, the 5G core should be deployed on a hyperscale cloud provider. This is to ensure scalability and dynamic orchestration of network resources over an infinitely scalable base. This blog post covers the key reasons to do so.
A brief recap of 5G Architecture Patterns
There are five possibilities depending on if the deployment is greenfield (brand new 5G network) versus brownfield (where there exists a footprint of 5G services). This was covered in the post – https://www.vamsitalkstech.com/5g/the-five-5g-architecture-solution-arch-patterns/
The first three options are reproduced below –
- The 5G Core and RAN are both deployed on the cloud provider – “Everything in the Cloud”
- The 5G Core runs on the cloud provider and the RAN on-premises on managed hardware – “Everything Managed but Hybrid”
- The 5G Core runs in the cloud and the RAN on-premises on self-managed servers – “Self Managed on-premise and hybrid”
Common theme & best practice – “Run the Core in the Cloud”
As one can see, the common theme to the first three architectures is that the Core runs on a cloud region at a hyperscaler. First off, this satisfies the requirement that the Core is deployed in a cloud-native environment – where different Network Functions ( NFs) can be scaled and orchestrated dynamically over a highly resilient platform architecture. This will obviate the need for the CSP to accommodate several expensive and high-effort services that can support the 5G Core.
The “5G Core network in a hyperscaler cloud region” deployment model is depicted in the below illustration. In this view, the entire mobile switching office (MSO) or central data center is deployed in the AWS region that also hosts OSS/BSS as well as the majority of the 5G Core. Managed hardware is then used for user-plane functions, such as UPF, RAN CU, and MEC.
Illustration – Reference architecture for 5G core control plane running in a cloud region
Architecture Benefits of Deploying the 5G Core in the cloud
- The key advantages of deploying the Core in the cloud are as follows –
- The ability to deploy on a flexible and extendable service-based network architecture
- Ready-made fault isolation across multiple datacenters. Hyperscalers like AWS provide multiple cloud Regions made up of individual AZs (Availability Zones) which are all isolated partitions. This allows CSPs to deploy network functions across separate data centers in logical groups (e.g an AMF set) which provide distributed high availability. Every AZ has its own redundant power, networking, and connectivity infrastructure that provides low latency networking. Moreover, Availability Zones are physically separated by a meaningful distance, many kilometers, from any other Availability Zones, although all are within 100 km (60 miles) of each other.
- Mix and match compute instances and core network configurations in different network slices for business reasons
- Allows for UE (User Equipment) to connect to the right user plane instances based on geographic criteria
- Support various patterns of high availability for the 5G core such as the 3GPP redundancy mechanisms – e.g pooling and NF-Set concept or generic geo-redundancy models by deploying each NF at every AZ. This will enable traffic re-routing (e.g from gNB to AMF) based on an AZ failure or load conditions.
- Performance optimization based on various criteria
- Leverage readily available DevOps pipelines for service updates – https://www.vamsitalkstech.com/5g/devops-for-5g-service-updates/
- Leverage IaC patterns for Day 0 setup – https://www.vamsitalkstech.com/5g/infrastructure-as-code-iac-for-5g-core-and-radio-network-setup/
- Cost optimization is based on obviating the need to overprovision capacity in a data center
- Security benefits as covered here – https://www.vamsitalkstech.com/kubernetes/how-does-kubernetes-security-in-the-cloud-work/
Conclusion
Running 5G Core in the cloud can result in manifold benefits as enumerated above. We will continue our discussion of 5G Reference Architectures. The next blog post will discuss the next architecture pattern “Everything Managed and Hybrid” where the 5G Core runs on the cloud provider and the RAN on-premises on managed hardware.
Featured Image by Harman Singh from Pixabay