Information Technology Reference
In-Depth Information
Figure 2-4 shows the components of fast switching. The following list outlines the process:
As a packet arrives from an inbound interface, a lookup is performed to determine if
a cached entry exists for the packet.
1
If none exists, the packet is process-switched.
2
Information gained from switching the first packet creates an entry in the fast cache.
3
If an entry already exists when the packet arrives, the packet is rewritten with the new
Layer 2 information for the outbound interface and is forwarded to that interface. The
packet is not copied to the system buffer, as in process switching.
4
Figure 2-4
IP Fast Switching
Process
Switching Path
Fast Cache
First
Packet
Processor
Packet
Memory
Inbound
Interface
Outbound
Interface
Optimum Switching
Optimum switching
stores the forwarding information and the MAC header rewrite
information in a 256-way radix tree. Using a 256-way tree reduces the number of steps that
must be taken when looking up a prefix, although more memory is needed. Optimum
switching is supported only on platforms based on the Route Switch Processor (RSP).
Distributed Optimum Switching
Distributed optimum switching
seeks to offload the packet-switching function from the
main CPU by moving the routing decision to the interface processors. This is possible only
on routing platforms that have dedicated CPUs per interface, such as Versatile Interface
Processors (VIPs). In the case of VIP, for example, the optimum cache is populated by the
RSP. When a packet is received, the VIP attempts to make the routing decision based on
that table. If the VIP can locate an entry on its local route cache, it switches the packet
without interrupting the RSP. If it fails, it enqueues the packet for the next configured
switching path (optimum switching, and then fast switching, and then process switching).
With distributed switching, access lists are copied to the VIPs, which allows the VIP to
check the packet against the access list without RSP intervention.
