We will cover a reference architecture for network slicing in this blogpost.
As we have seen in other blogs, Network slicing really enables the creation of multiple virtual networks on a single physical network(s). enables the definition of multiple virtual networks on single physical networking infrastructure. A network slice represents an independent virtualized instance defined by the allocation of a subset of the available network resources. Typically, network slices are tailored to meet specific requirements of a set of applications and services. Network slicing consists of defining an isolated subset of the available virtual resources (computation, networking, storage) and a set of rules for identifying the traffic that will run on those. A network slice consists of a set of virtual resources and the traffic flows associated with it.
The resources used by a Network Slice include –
- Bandwidth
- Topology of the network that it is hosted on
- CPU resources
- Storage
- Control plane resources and forwarding tables etc
- Traffic for isolation from other slices in a multi-tenant deployment
Leveraging the above building blocks, a network slice can be created dynamically and be defined based on a subset of resources. For instance, in a typical 5G network, slices of RAN (Radio Access Networks) can be created while enabling controlling infrastructure to define the provisioning of the network, compute, and storage resources that make up the slice. Network slicing thus builds up SDN (Software Defined Networking) and NFV (Network Functions Virtualization) concepts that are widely used across the telco industry.
Let us now discuss a reference architecture as proposed by the 3GPP.
3GPP identifies the three core layers in a network slicing implementation:
- Service Instance Layer: The Service Instance Layer is the main hosting provider that provides services or applications provided to the end-user
- Network Slice Instance Layer: A Network Slice Instance represents a collection of Network functions (NFs) from infrastructure layers that enable the formation of a network slice.
- Resource Layer: The Resource Layer provides resources called subnetwork instances which represent every kind of resource needed by a network slice – network, computation, storage etc. These are typically represented as collections of Virtual Network Functions (VNFs) or Physical Network Functions (PNFs).
- Network Slice Controller This is the overall orchestrator that controls the lifecycle of the slice (creation, provisioning, deployment, monitoring, and deletion). Vendors are free to define this based on the value addition they want to offer their customers.
Conclusion
The next set of blogs will dive into real-world network architectures of some of the elements we have been discussing in blogs this year – 5G Core, RAN, and Network Slicing. We will use AWS (Amazon Web Services) as the host for their deployments.
Image credit: Ivan Torres