cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
4578
Views
3
Helpful
3
Replies

What is the difference between fast switching, process switching and CEF?

vijayendran
Level 1
Level 1

Hi All

Could you explain me the difference between Fast Switching, Process Switching and CEF?

I did read Cisco IOS Swtching Path Overview document available on cisco.com. But could not understand.

Thanks and regards

Vijayendran Sridharan

2 Accepted Solutions

Accepted Solutions

Hello Vijayendran,

check out this document, it explains the differences in detail:

How to Choose the Best Router Switching Path for Your Network

http://www.cisco.com/en/US/partner/tech/tk827/tk831/technologies_white_paper09186a00800a62d9.shtml#topic1

Regards,

Georg

View solution in original post

Basically, wth process switching, every packet entering an interface creates a CPU interrupt, and processed by the main CPU, with fast switching, one step forward: every packet to a certain destination is process switched first and a table called fast switching cache is formed. This table consists of: a destination IP, associated layer-2 address, and output interface. This mechanism makes pavket switcing process faster. A packet to a certain destination is process switched first and meanwhile associated fast switching cache entry is formed. Subsequent packets are fast switched, that's to say: no need for the main CPU for route lookup, fast sw cache contains enough info. CEF is the latest step in switching technology, it has solutions to some drawbacks of fast switching: Fats switching cache is file as traffic comes in and fast sw cache consists of single host entries. CEF maintains two tables: Adjacency Table and FIB (forwarding information base). FIB contains destination networks (learnt from static or routing protocols) and associated outgoing interfaces. Adjacency table contains layer-2 addresses for layer-3 next-hops in FIB. FIB is not traffic driven as with fast switching cache, it's formed as soon as route table is filled, and contains network entries contrast to fast switching cache. CEF and it's distributed form (dCEF) are the most efficient packet switching mechanisms for Cisco routers. Also CEF is required for certain QoS applications.

Regards.

View solution in original post

3 Replies 3

Hello Vijayendran,

check out this document, it explains the differences in detail:

How to Choose the Best Router Switching Path for Your Network

http://www.cisco.com/en/US/partner/tech/tk827/tk831/technologies_white_paper09186a00800a62d9.shtml#topic1

Regards,

Georg

Basically, wth process switching, every packet entering an interface creates a CPU interrupt, and processed by the main CPU, with fast switching, one step forward: every packet to a certain destination is process switched first and a table called fast switching cache is formed. This table consists of: a destination IP, associated layer-2 address, and output interface. This mechanism makes pavket switcing process faster. A packet to a certain destination is process switched first and meanwhile associated fast switching cache entry is formed. Subsequent packets are fast switched, that's to say: no need for the main CPU for route lookup, fast sw cache contains enough info. CEF is the latest step in switching technology, it has solutions to some drawbacks of fast switching: Fats switching cache is file as traffic comes in and fast sw cache consists of single host entries. CEF maintains two tables: Adjacency Table and FIB (forwarding information base). FIB contains destination networks (learnt from static or routing protocols) and associated outgoing interfaces. Adjacency table contains layer-2 addresses for layer-3 next-hops in FIB. FIB is not traffic driven as with fast switching cache, it's formed as soon as route table is filled, and contains network entries contrast to fast switching cache. CEF and it's distributed form (dCEF) are the most efficient packet switching mechanisms for Cisco routers. Also CEF is required for certain QoS applications.

Regards.

Actually, process switched packets are handled by a process on the router, ip input, so the steps are:

-- packet received and placed on the receive ring of the interface

-- CPU interrupted, packet classified and moved to the ip input queue

-- ip input scheduled, eventually runs, does the correct lookups, and then rewrites the mac header

-- packet is placed on the output queue of the right interface

With fast, if there is a fast cache entry, the packet is switched in the first interrupt, after the packet is received. If there is no fast cache entry, the packet is process switched. With cef, the packet is processed in the first interrupt, or it is "punted" to process switching.

Take a look at Inside Cisco IOS Software.

:-)

Russ.W