Web Analytics
Skip to content

Network Fabric Router

Network Functions Virtualisation#

The Network Fabric Router (NFR) is a platform built from a range of virtualised network functions (NFV).

The network fabric (INF) is instantiated on these software instances running on bare metal x86 servers and the following network functions are then chosen depening on the scope of the solutions being deployed:

  • Dynamic routing infrastructure
  • Packet vectoring engine
  • Bandwidth aggregation nodes
  • Bi-directional QoS engine
  • Software defined cloud interconnect
  • ZTNA transit gateway
  • Site to site transit gateway
  • Real-time monitoring instances
  • Telemetry collection nodes

The core for the EVX edge#

Where EVX appliances (phsyical or virtual) reside in branch offices and remote sites, NFRs form the network backbone that they each connect to.

Every EVX is logically connected to a minimum of 2x NFR instances (by default across 2 geographically diverse datacentes).

These NFRs act as the decentralised transit hubs, as well as performing the vital functions of the network fabric above.

Decentralised network backbones#

Deployed in configurations spanning multiple datacentre locations, NFRs work together, but independantly within a WAN solution.

Configurations are pushed to NFR nodes when changes are required, but there is no dependency on any orchestration systems to continue to operate or launch. If devices lose power, they will boot up and restore their configuration without any need for external support.

NFRs usually operate in clusters of 2 or 3 across the same number of datacentres. Connections are usually split evenly across the cluster in a spine and leaf topology, allowing all but 1 NFR to go down while maintain full network capabilities for the WANs they host.

Bare metal#

While NFR can run as a virtual server, better and more reliable performance is achieved with bare metal servers.

NFR is a processer heavy application, and very light on RAM and disk. Packets are processed at wire speed, and the faster/more capable the processor, the more bandwidth and more Connections can be supported by any individual instance.