[nsp] High latency

Kevin kevin at honeycomb.net
Wed Nov 12 15:52:32 EST 2003


Ok, here's another question to reduce traffic load,
Most of my traffic from on internal subnet gets route-mapped
To one of my external connections, is it possible to bypass the route
table with a route-map
It seems like even though the packet is route-mapped, it still has to
scan through the route table to see where it goes?

Kevin, 

-----Original Message-----
From: Haesu [mailto:haesu at towardex.com] 
Sent: Wednesday, November 12, 2003 1:44 PM
To: Kevin; cisco-nsp at puck.nether.net
Subject: Re: [nsp] High latency


Do you use VIP's (most likely yes) with enough RAM in them?
If so, make sure distributed-CEF is enabled on, as it will reduce the
load on the RSP quite a bit with ip input

-hc

-- 
Haesu C.
TowardEX Technologies, Inc.
Consulting, colocation, web hosting, network design and implementation
http://www.towardex.com | haesu at towardex.com
Cell: (978)394-2867     | Office: (978)263-3399 Ext. 170
Fax: (978)263-0033      | POC: HAESU-ARIN

On Wed, Nov 12, 2003 at 01:20:07PM -0600, Kevin wrote:
> Yes I'm running BGP on the router with full routes, here's the output 
> from "sh proc cpu hist" I am running CEF on the router also.
> 
> border1#show processes cpu history
> 
>      3344444888882222233333333332222222222333333333322222222222
>      0011111777775555500000111119999977777000007777766666666664
> 100
>  90         *****
>  80         *****
>  70         *****
>  60         *****
>  50         *****
>  40    **********                              *****
>  30  *********************************************************
>  20  **********************************************************
>  10  **********************************************************
>     0....5....1....1....2....2....3....3....4....4....5....5....
>               0    5    0    5    0    5    0    5    0    5
> 
>                CPU% per second (last 60 seconds)
> 
> 
>      9999999998888777785999877899987789998899999887899889999878
>      2232488961410343713091098769147741862036888736488233800094
> 100       ****           *     **      **   ****    **   *
>  90  *********          ***   ****    ***  ******   **  ****
>  80  *************   ** ***************************************
>  70  ****************** ***************************************
>  60  ****************** ***************************************
>  50  **************#*****###****#*****#*####*****#*######******
>  40  ****#******###############################################
>  30  ##########################################################
>  20  ##########################################################
>  10  ##########################################################
>     0....5....1....1....2....2....3....3....4....4....5....5....
>               0    5    0    5    0    5    0    5    0    5
> 
>                CPU% per minute (last 60 minutes)
>               * = maximum CPU%   # = average CPU%
> 
> It does look like bgp scanner is running at the inteveral, How do I 
> reduce the IP Input load?
> 
> Kevin,
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net 
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Streiner, 
> Justin
> Sent: Wednesday, November 12, 2003 12:54 PM
> To: cisco-nsp at puck.nether.net
> Subject: Re: [nsp] High latency
> 
> 
> On Wed, 12 Nov 2003, Kevin wrote:
> 
> > I'm having a really strange Issue with my 7513,
> > It seems like the 7513 will make the latency on any link Jump up to 
> > a peak of 100ms to 400ms for a few packets then settle down
> 
> > to below 1ms Then after a little while (30sec to 1min) it'll do the
> > same thing, The cpu load when this happens is around 30%-40% (this 
> > load is constant during the day),
> > None of the interfaces are overloaded, it happens on all the
> interfaces.
> > Is there anything else I can do to diagnose the problem? It's a 7513

> > with an RSP4 and 12.2(14)s3.
> 
> Are you running BGP on this router?  If so, during these times of high

> latency, does the "BGP Scanner" process tear up lots of CPU when you 
> look at a "sho proc cpu"?
> 
> If you do a "sho proc cpu hist", do you see lots of regular spikes up 
> to around 100%?
> 
> If so, I'll bet what you're seeing is the BGP Scanner process walking 
> the RIB.  If your router is heavily loaded and/or you're taking 
> multiple full feeds this process can eat up lots of CPU and produce 
> the latency you're seeing.
> 
> Running CEF won't specifically cure this problem, but it can help 
> reduce the amount of CPU your router spends on other tasks, 
> particularly IP Input.
> 
> jms
> _______________________________________________
> 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/







More information about the cisco-nsp mailing list