>Stuart,
>
>I do not think any packets could be switched without CPU. Fast switching
>needs CPU, it just does not need the CPU to call the ip input process. If
>the router needs to search the routing table, it will need ip input process,
>other than that, the CPU can use cache during one interrupt. As you
>mentioned there are different cache, it all depends on how the router creats
>them.
>Just my 2 cents.
>
>Thanks
>Jack
Distributed CEF, distributed NetFlow, silicon, autonomous, and
arguably optimum switching get at least some of the forwarding out of
the main processor. Typically, the first packet to a destination goes
through the main processor, and creates a cache entry that will be
used for subsequent packets.
dCEF is a little different, in that it creates FIBs that mirror the
RIB, so, once routing protocols or other mechanisms put a destination
into the RIB, the FIBs change in response. There is no need to wait
for a first packet.
>
>
>""Stuart Potts"" <[EMAIL PROTECTED]> wrote in message
>[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
>> Hi,
>>
>> Brief:
>>
>> Fast switching is as follows, 1st packet is processed switched, 2 packet
>to
> > destination is fast switched. i.e. it never goes through the cpu.
The fast switched packets do go through the CPU, but their
destination is looked up in a cache that is faster for lookup than
the main routing table. Fast switching, for other reasons, do reduce
the CPU load.
> >
>> L3 switching , MLS, Multi layer switching, This is based on traffic flows,
>> ie we could swich on a certain tcp flow, or certain source/detination and
>> destination port. as defined by access lists.
>>
>> The three flow masks are as follows:
>>
>> destination-ipóThe least-specific flow mask. The MLS-SE maintains one MLS
>> entry for each destination IP address. All flows to a given destination IP
>> address use this MLS entry. This mode is used if there are no access lists
>> configured on any of the MLS-RP interfaces.
>>
>> source-destination-ipóThe MLS-SE maintains one MLS entry for each source
>and
>> destination IP address pair. All flows between a given source and
>> destination use this MLS entry regardless of the IP protocol ports. This
>> mode is used if there is a standard access list on any of the MLS-RP
>> interfaces.
>>
>> ip-flowóThe most-specific flow mask. The MLS-SE creates and maintains a
>> separate MLS cache entry for every IP flow. An ip-flow entry includes the
>> source IP address, destination IP address, protocol, and protocol ports.
>> This mode is used if there is an extended access list on any of the MLS-RP
>> interfaces.
>>
>>
>>
>>
>http://www.cisco.com/univercd/cc/td/doc/product/lan/cat5000/rel_5_2/layer3/m
>> ls.htm#xtocid171391
>>
>> http://www.cisco.com/warp/customer/473/55.html
>>
> > /Stuart.
>>
> >
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]