Re: Policing shaping for ingress/egress

From: Ccieyarub <ccieyarub_at_gmail.com>
Date: Wed, 8 Dec 2010 12:36:32 -0500

Guys what's your thoughts on that one

Cheers
Bert

On 2010-12-08, at 12:29 AM, Ccieyarub <ccieyarub_at_gmail.com> wrote:

> Hi all
> Sorry if my mail is too long
> In a network
> Small switch R0 -----Access switch R1 -------mvrf lite routerR2
> ------core mpls/ l3vpn network
> for the port between R1 and R2 there is a requirement to pass only
> 100meg traffic as R0 to R1 link is only 100 Meg
>
> The solution that was provided is to apply policing on the ingress
> and egress of the interface R2 connected to R1 with the following
> policy
> policy-map XX
> class class-default
> police 100000000 50000 50000 conform-action transmit
> exceed-action drop violate-action drop
> !
>
> another suggestion is to apply policing on ingress and shaping on
> egress. Which i would more agree with because there is lots of tcp
> traffic and policing is not good for tcp. So we would like to use
> the policing policy above on the ingress of the interface only
> however in the outbound/egress We would like to use
>
> policy-map YY_100M
> class class-default
> shape peak 100000000
>
>
> Question
> - which solution is better using policing on both ingress and egress
> or using policing on ingress with shaping on egress !??
> - if we used shaping on egress are we introducing big delays in
> this case due to queuing for shaping with default queue
> - how big is the effect of policing with tcp window size if used
> policing on egress
>
> Thanks
> Bert

Blogs and organic groups at http://www.ccie.net
Received on Wed Dec 08 2010 - 12:36:32 ART

This archive was generated by hypermail 2.2.0 : Sat Jan 01 2011 - 09:37:49 ART