Tuesday, 17 January 2012

Cisco IP Interface switching

Whilst studying some text on NAT, I kept noticing segements of text referring to Cisco switching mechanisms.

Here's a line of the text:
"On a NAT tranlsation table, an asterix means that the translation is occurring in the fast-switched path.  By default the first packet in a NAT translation will always be process-switched" - so whats all this fast/process switched business?

Cisco layer 3 devices have three switching modes, Process Switching, Fast Switching, and Cisco Express Forwarding switching.

If you see the configuration line "'no ip route-cache" on an interface the packets entering the interface will be process switched, which means the CPU will do the switching - a potential burden on a router.
Fast switching overcame the issue by route caching the first packet so that others in the same flow didn't have to hit the CPU - these routes were housed in hardware lookup tables that are independent of the CPU.
CEF (Cisco Express Forwarding) builds on fast switching by using FIB (forwarding information bases) and adjacency tables to quickly link packets/flows with routing table entries and adjacent devices.

This is one way of determining what each interface is doing, have a look for the commands:

*Nothing specified in the interface?*
CEF usually enabled by default - you may see ip cef listed earlier in your config
No ip route-cache
Process switched
ip route-cache
Fast switched

That seems to make a bit more sense.

No comments:

Post a Comment