[c-nsp] C3560X Layer 3 throughput

[Nick Cutting]

Oh a 3560 not 3650 ;)

Check every interface is running CEF.

sh ip interface | in IP route-cache

I had to debug ip packet detail on a 3750 recently, to see what was being sent to the CPU - I tracked it down to IP options being sent through the switch that were being punted to the CPU. You will see source and destinations.

I used this guide:

http://www.cisco.com/c/en/us/support/docs/switches/catalyst-3750-series-switches/68461-high-cpu-utilization-cat3750.html

Now most people are afraid of running debug ip packet detail on a production device.  I am too.
However, this is not a software ISR router, only control plane traffic and punted packets "should" show up. 

....provided you are definitely running fast switching / CEF




-----Original Message-----
From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of David Wilkinson
Sent: 16 February 2016 12:59
To: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] C3560X Layer 3 throughput

Phil, That is what I expected as well.

Running the universal image with the ipservices license.

Nick, Nothing facny like tunnels, very basic config. running IOS, not XE


Cisco IOS Software, C3560E Software (C3560E-UNIVERSALK9-M), Version 15.0(2)SE9, RELEASE SOFTWARE (fc1) Technical Support: http://www.cisco.com/techsupport Copyright (c) 1986-2015 by Cisco Systems, Inc.
Compiled Tue 01-Dec-15 06:42 by prod_rel_team

ROM: Bootstrap program is C3560E boot loader
BOOTLDR: C3560E Boot Loader (C3560X-HBOOT-M) Version 12.2(53r)SE2, RELEASE SOFTWARE (fc1)

cisco WS-C3560X-24 (PowerPC405) processor (revision K0) with 262144K bytes of memory.

Last reset from power-on
1 Virtual Ethernet interface
1 FastEthernet interface
28 Gigabit Ethernet interfaces
2 Ten Gigabit Ethernet interfaces


Regards

David


On 16/02/2016 12:28, Elmar K. Bins wrote:
> p.mayers at imperial.ac.uk (Phil Mayers) wrote:
>
>>> What is the real world expected throughput on the C3560X devices 
>>> when doing basic Layer 3 routing?
>> I'd expect it to be able to do line-rate with almost no CPU - it 
>> should be forwarding in hardware.
> That was my experience with all Cisco L3 switches (35x, 375x, et al)
> - as long as nothing gets punted to CPU (due to, e.g., netflow f***ups 
> etc.), forwarding is line rate, meaning your throughput depends on the 
> size of your packets.
>
>
>> Check you have the IP rather than LAN image running, that you've got 
>> an appropriate SDM template for routing, that you haven't got any 
>> unsupported features enabled, and that you aren't otherwise inducing 
>> it to punt to CPU e.g. by getting a lot of IP redirects/ICMP unreach or other such activity.
> The SDM template should not have too much of an influence, unless you 
> are constantly timing out stuff.
>
> El.
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net 
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/

_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/