Re: Policing or Shaping

From: CCIE To Be (ccie.tobe81@gmail.com)
Date: Tue Jan 20 2009 - 13:12:34 ARST


So, after this discussion I reached on the following solutions,

police cir percent 75 conform-action transmit exceed-action
set-frde-transmit using MQC

OR

shape average percent 75 ( using MQC )
random-detect
random-detect ecn ( becasue FECN/BECN not allowed )

HTH,

On Tue, Jan 20, 2009 at 4:59 PM, Pavel Bykov <slidersv@gmail.com> wrote:

> Uh...
> Net He,
> that's not how interfaces work.
> You cannot send at any other speed then the interface speed. Interface can
> have either on or off state - don't forget that it's digital.
> Check out chapter on test setup in the following document to understand the
> need for queueing:
> http://www.boxoid.org/cisco/MAX-RESERVED-BANDWIDTH-AND-CBWFQ.pdf
> 75% utilization is counted using packets over time and with such high
> utilization, there is 99% chance that router is queuing like crazy. The only
> scenario this would not be true is if you had shapers in the path before or
> if you'd use real time operating systems.
>
>
> Raul: Policing can mark/remark packets that exceed rate, so policing with
> "set-de-transmit" as exceed action could work for you. But to tell the other
> router what to do... um... You could write a complecated EEM script. Or
> maybe just use something like "frame-relay congestion threshold de 75"? Can
> you use foresight messages?
> Telling othber routers what to do without FECN/BECN could be complicated.
> You could tell end-hosts what to do, by dropping packets...
>
>
>
>
> On Tue, Jan 20, 2009 at 3:12 AM, NET HE <he_net@hotmail.com> wrote:
>
>> Since the condition is 70% utilization of bandwidth, in other words, there
>> is no congestion. So congestion management of using queues can't be used.
>> And congestion avoidance works on queues instead of bandwidth utilization,
>> so congestion avoidance is not an option either.
>>
>> Traffic shaping can't set bits to inform other network nodes, so it is
>> excluded.
>>
>> Traffic policing can set bits based on bandwidth utilization, but the bits
>> are to inform downstream network instead of upstream network.
>>
>> Best Regards,
>> Net (Xin) He
>>
>>
>>
>>
>> > Date: Mon, 19 Jan 2009 19:32:16 +0400
>> > Subject: Re: Policing or Shaping
>> > From: ccie.tobe81@gmail.com
>> > To: sadiqtanko@gmail.com
>> > CC: slidersv@gmail.com; ccielab@groupstudy.com
>>
>> >
>> > Hi,
>> >
>> > Local router needs to inform remote router that network is congested and
>> > remote router needs to slow down packet forwarding once traffic
>> utilization
>> > on FR interface of local router will reach 75%. Don't use FECN or BECN.
>> >
>> >
>> > HTH
>> >
>> > On Mon, Jan 19, 2009 at 7:16 PM, Sadiq Yakasai <sadiqtanko@gmail.com>
>> wrote:
>> >
>> > > Well, cos of "and we need to inform remote router that there is
>> congestion
>> > > in network.", I would tend to think we need to inform the remote
>> router that
>> > > there is congestion on the network, so that an immediate action would
>> be
>> > > taken.
>> > >
>> > > I think there isnt enough information from the question to provide an
>> > > exactly solution.
>> > >
>> > > Anyways, just my 2 cents.
>> > >
>> > > Sadiq
>> >
>> >
>> > Blogs and organic groups at http://www.ccie.net
>> >
>> > _______________________________________________________________________
>> > Subscription information may be found at:
>> > http://www.groupstudy.com/list/CCIELab.html
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>>
>>
>> ------------------------------
>> Share your holiday memories for free with Windows LiveT Photos. Get
>> started now. <http://www.microsoft.com/windows/windowslive/photos.aspx>
>>
>
>
>
> --
> Pavel Bykov
> ----------------
> Don't forget to help stopping the braindumps, use of which reduces value of
> your certifications. Sign the petition at http://www.stopbraindumps.com/

Blogs and organic groups at http://www.ccie.net



This archive was generated by hypermail 2.1.4 : Sun Mar 01 2009 - 09:43:39 ARST