To optimize the Edge deployment to run workloads without significant management overhead, the management plane functionality for the Edges is usually implemented in a site remote to the Edge such as Core data center. While this imposes some constraints on networking (including link availability and end to end latency), this model of remote management is useful for constrained environments such as Telco Edges.

The benefit of this model is the ease and simplicity with which the entire Telco infrastructure can be administered. Instead of connecting to each site to configure and control the resources at that site, users can access the centralized management at the Core data center, which can give them access to all the Edge sites under its purview.

In some deployments, the number of Telco Edges can be very large, running into several thousands. In such cases, a hierarchical management approach is optimal. Irrespective of the model of aggregation, an architectural principle that an edge site is managed from a central site is used. This reference architecture considers a model where a group of sites, both Telco Edges and Aggregation sites, are managed from a single management instance at a core site. This is depicted in the following figure:

Figure 1. Telco Edge Reference Model
Telco Edge Reference Model

The Telco Edge and Aggregation sites are collapsed into a single Telco Edge that is managed from a Core data center. We use the term "region" to indicate a group of Edge sites. The preceding figure depicts the components of a Telco edge representing the NFV infrastructure that includes compute, networking, and storage.

The Core site is connected to the Edge site through a Telco network generically described as metro/WAN in this reference architecture. Examples of such networks include Metro Ethernet, MPLS, and so on. However, the actual technology used is not pertinent to this reference architecture. Because the core sites and the Telco Edges are connected over Layer 3, it is important that a routed network topology exists between the sites.

Layer 2 networks are expected to be terminated at the Provider Edge (PE) routers at the Core and Edge data centers. A Layer 3 path and connection between the core PE router and the Edge PE router through the metro or WAN network is assumed to be configured already. It is important to ensure that sufficient bandwidth at low latency is ensured for this Core site to Edge site connectivity.