Users Guide

Conventional Core

A conventional core is a three-tier network that is typically chassis-based and is composed of the

following:

• Core — The core layer routes traffic to and from the Internet and the extranet. High availability, which

provides redundancy and resiliency, requires chassis-based core routers.

• Aggregation layer — The aggregation layer connects with top of rack (ToR) switches and aggregates

the traffic into fewer high-density interfaces such as 10GbE or 40GbE. This layer aggregates the traffic

to the core layer.

• Access layer (ToR) — The access layer typically contains ToRs. A ToR is a small form-factor switch

that sits on top of the rack and allows all the servers in the rack to be cabled into the switch. A ToR

has a small 1–2 rack unit (RU) form factor.

Figure 3. Conventional Core

Distributed Core

A distributed core is a two-tier architecture composed of multiple interconnected switches, providing a

scalable, high-performance network that replaces the traditional and aggregation layers in a conventional

core. Switches are arranged as spines and leaves. The spines in the fabric connect the leaves using a

routing protocol. The leaves’ edge ports connect to the switches, ToR switches, servers, other devices,

and the WAN. The spines move traffic bidirectionally between the leaves to provide redundancy and load

balancing. Collectively, the spine and leaf architecture forms the distribute core fabric.

This two-tier network design allows traffic to move more efficiently in the core and at a higher bandwidth

with lower latencies than most traditional three-tier networks. Since there is no single point of failure that

can disrupt the entire fabric, the distributed core architecture is more resilient and there is less impact on

the network if a link or node failure occurs. AFM views the distributed core as one logical switch.

Supported Fabric Types