[c-nsp] rate limiting newbie question

[Erdem Sener]

based on commands supported by the ios, you could use 'rate-limit'
command on the interface along with the policy-map as well. (I'm not
sure if this is the best practice, though)

a rate-limit command should look like this:

rate-limit input|output <access-group xxx> bw1 normal_burst
maximum_burst conform-action <action> exceed-action <action>

for numeric values, the algorithm below works fine with almost all ios versions:

normal_burst= (bw1 / 8) * 1.5
maximum_burst = 2 x normal_burst

For <action> options, you can always use '?' for help

HTH

On 4/13/06, Oliver Boehmer (oboehmer) <oboehmer at cisco.com> wrote:
> Fredrik.Jacobsson at enskilda.se <> wrote on Thursday, April 13, 2006 2:04
> PM:
>
> > Hi!
> >
> > I (customer) have a ip-vpn service provided dropping off an ethernet
> > interface.
> > The service is for 8 Mbit/s.
> >
> > Obviously I want to send my packets in a controlled behaviour and do
> > queuing myself to prioritize some packets in front of other so no
> > congestion
> > occurs at the service provider routers.
> >
> > Today I have this on my old link, a serial interface: (on a
> > 2800-router
> > running 12.4)
> >
> > policy-map mypolicy
> >  class voip
> >   pri 72
> >  class business-critical
> >   band 1024
> >  class class-default
> >   fair-queue
> >
> > Now I also need to make sure that the maximum bandwidth for entire
> > policy doesn't exceed 8192 kbps.
> > Do I really need to set a shape/police for each class? I don't want
> > to limit a class...but the entire interface speed.
>
> policy-map parent
>  shape average 8192000
>  service-policy mypolicy
> !
> int fastethernet0
>  service-policy output parent
>
> will do the trick..
>
>        oli
>
> P.S: Check the archives, this topic has come up quite frequently
> lately..
>
> _______________________________________________
> 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/
>