All PeerEdge® Fabric Facilities are interconnected using highly redundant private fiber links. Each Fabric Region is privately peered to 1000+ service provider networks and all global public cloud providers.

These interconnections allow The Fabric to be physically extended to remote locations using Fabric Local Loops without ever touching the public internet.

The Fabric is interconnected over the Internet using Voice Orchestrators or Network Orchestrators. These Orchestrators can be deployed inside existing enterprise environments, Managed Hosting or Managed Edge Devices.

One of the key limitations for software-based routing stacks is OS kernel-level packet processing. Standard operating systems aren’t designed to process millions of packets per second or hundreds of gigabytes of throughput. This has historically been the territory of ASICs and specialised hardware, generally dominated by commercial network vendors. With the advent of VPP for packet processing and DPDK for direct access to network devices, OS-based limitations have been removed. Both VPP and DPDK are heavily used inside PeerEdge® Data .

All traffic exchanged between Network Fabric Extenders and The Fabric are encrypted using AES- based algorithms, which is accelerated using AES-NI instruction sets. Modern processor accelerate encryption workloads, without requiring a hardware encryption engine. Most cloud providers support AES-NI instruction sets.

Routing information is exchanged using Multiprotocol BGP (MP-BGP), which treats MAC Addresses as routes. This loop-free architecture, eliminates spanning tree protocol (STP) and ends broadcast storms (because there is no broadcast). MAC flapping detection and mitigation prevents unicast and multicast loops in an all-active multihomed topology.

PeerEdge® Network Orchestrators support edge-device multihoming as well as MAC address mobility and load balancing across dual-active links. When split between Orchestrators, the links provide true HA failover at the network edge.