Omni Switch/Router Distributed Switching Fabric
Page 1-3

Omni Switch/Router Distributed Switching Fabric

Many switches in the market employ a shared memory architecture, which uses a central
switching engine to send data to the appropriate port. As shown in the figure below, data
enters the input port ( below), crosses the switching fabric on its way to the central switch-
ing engine( below), and again crosses the switching fabric ( below) before exiting the
appropriate output port ( below).
Traditional Shared Memory Architecture
In contrast, Omni Switch/Router switches use a distributed switching fabric. As shown in the
figure below, data enters the input port and crosses the switching fabric only once before
exiting the appropriate output port. Compared to the shared memory architecture, only half
as much bandwidth is required since data just crosses the switching fabric once.
Omni Switch/Router Distributed Switching Fabric
Switch Fabric
Central Switching Engine
I/O Port I/O Port
Note: Unlike distributed switch
fabric, data is forced to cross
switch fabric twice.
➋➌
Switch Fabric
I/O Port I/O Port