[e-nsp] Extreme Switch CPU utilization on port-mirroring ? (Hamid Ahmed)

David Newman dnewman at networktest.com
Mon Aug 13 16:01:00 EDT 2007


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 8/13/07 12:05 PM, Stéphane Grosjean wrote:
> Hi,
> 
> 
>> I have an extreme switch Summit 400-24p. I am running
>> port-mirroring for >two ports on tthe switch. I need to check what
>> impact does it make on the >CPU when i run port mirroring for a
>> said number of ports?
> 
> Extreme switches are asic based, not cpu. That means there's no real
> big deal on cpu usage. I assume it should be just fine if you stay on
> the official recommendations.

All switches, even Extreme's, use CPUs to "slow-path" certain traffic
such as control-plane messages (no one does routing purely in hardware)
and data-plane packets with IP options set.

I have not seen any performance degradation from port mirroring but then
again I have not attempted to measure this under heavy load. I would be
curious to know what performance costs, if any, exist for mirroring.

dn



-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (Darwin)

iD8DBQFGwLh8yPxGVjntI4IRAlHEAKDEENXLGHfrzV17yjb9EB6uXQfD7ACgl0Oy
JcsVuiCBklxnClmJ1BZ4I7w=
=SZFm
-----END PGP SIGNATURE-----


More information about the extreme-nsp mailing list