We covered an introduction to network slicing architecture here – https://www.vamsitalkstech.com/5g/the-systems-architecture-of-network-slicing/. Let us now understand how a software-based stack running on Containers would support a layered deployment of the above. We are approaching 5G and slicing from a software architect and solution architecture standpoint as opposed to a purely telco-centric look.
A network slice is used by tenants to rent or leverage end-to-end connectivity from a network provider. This enables tenants to run their own private 5G services spanning the 3 major types – https://www.vamsitalkstech.com/cloud/the-three-key-use-case-areas-for-5g/ of services namely – eMBB, URLLC, and mMTC. Network slicing can be enabled across all three types of 5G services.
- Building up from the bottom is the resource layer. This provides all manner of infrastructure – compute (VMs/Containers/Bare metal instances etc), networking, storage, and an ability to onboard NFs using an orchestrator. This again spans both DU/CU and the 5G core itself. The resource layer can be hosted on a cloud or on-prem or using a mix and match the architecture. The resource layer also provides a containerization platform, such as Amazon EKS.
- The network slicing layer is where most of the action is. It builds on the Resource Layer and builds on dedicated infrastructure per slice. Slices can be created in the RAN or Core or even in the Transport network as long as the NFs themselves are enabled to be slice compliant. This means that functions such as UPF or gNB-DU/CU etc. all can be sliced. What this means in practical terms is – when implemented on say AWS, this layer has full integration with various services such as CodeFormation, RDS, EBS, Lambda, and DevOps tooling such as CodePipine (to be covered in the next blog). The slice layer also has the ability to fully manage NS (network slice) lifecycle as discussed in this blog – https://www.vamsitalkstech.com/5g/the-lifecycle-of-a-network-slice/
- The Service support layer brings the full benefit of Cloudification to slicing and includes a range of support functions ranging from Caching, Service Orchestration (optional capability based on the vendor), load balancing, workflow etc. Each slice runs various NF instances and one has the optionality of bringing in a Service Mesh for inter-service communication
- The Service instance layer interfaces directly with clients and business users that request slices of the network as a service.
While Network Slicing is undoubtedly key in enabling operators to monetize 5G, the principles of DevOps as applied to network slicing and NF updates are key in helping achieve a competitive edge.
The next blogpost will discuss Network Function and Network Slice service update automation using DevOps pipelines.
Image source: https://unsplash.com/photos/QdAAasrZhdk