From: NGUYEN Trung (trung@vnsystem.net)
Date: Sun Apr 03 2005 - 05:46:07 GMT-3
Hi Tim.
Just an error typing.
You know it ;)
N.Trung
----- Original Message -----
From: "ccie2be" <ccie2be@nyc.rr.com>
To: "'Dennis J. Hartmann'" <dennisjhartmann@hotmail.com>; "Group Study"
<ccielab@groupstudy.com>
Sent: Thursday, March 31, 2005 5:31 PM
Subject: RE: Traffic shaping question #1
> Dennis,
>
> Your narrative doesn't match up with your example.
>
> In the 1st line of narrative, you use a plus sign (+) but in your example,
> you show multiplication.
>
> Which one is correct? Hmmmm
>
> Tim
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> Dennis J. Hartmann
> Sent: Wednesday, March 30, 2005 10:47 PM
> To: 'Chris'; 'Jongsoo kim'; 'Sundar Palaniappan'
> Cc: 'Group Study'
> Subject: RE: Traffic shaping question #1
>
> Chris' understanding is correct. Shape peak will allow bursting of
> CIR + (1 + Bc/Be), while average will constrain the rate to CIR. The
> default Bc and Be is 8000 bytes if using the shape peak option which will
> allow you to double your data rate via bursting capabilities.
>
> Shape peak 32000
>
> 32000 * (1 + 8000/8000)
> 32000 * 2
> 64000 (effective throughput)
>
> The shape average is constrained to CIR.
>
> -Dennis Hartmann
> White Pine Communications
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> Chris
> Sent: Wednesday, March 30, 2005 9:09 PM
> To: 'Jongsoo kim'; 'Sundar Palaniappan'
> Cc: 'Group Study'
> Subject: RE: Traffic shaping question #1
>
> My understanding is that traffic shaping under a policy map shapes the
> traffic coming out of a queue or set of queues if you are nesting classes
> in
> your policy map. Classified traffic can get more then the configured
> bandwidth and shaping average will constrain it to the cir. Because the
> shaper is after the queues it is active during congestion. That is my
> understanding. That is why something like the config below would work. It
> would give data1 a min of 32k, data2 a min of 64k but shape both classes
> and
> any other traffic to 128 k max. This is my understanding.
>
>
>
> policy-map test
> class data1
> bandwidth 32
> class data2
> bandwidth 64
>
> policy-map shapeit
> class class-default
> shape average 128000
> service-policy test
>
>
>
>
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> Jongsoo kim
> Sent: Wednesday, March 30, 2005 6:38 PM
> To: Sundar Palaniappan
> Cc: Group Study
> Subject: Re: Traffic shaping question #1
>
> Based on my recent study, "shape averagre or peak" only deals with CIR
> upper boundary. "bandwidth" comand deals with CIR lower boundary.
> In case of Frame-relay, "shape adaptive" command can be used instead of
> "bandwdith"
>
> traffic shaping under policy map is performed when there is no congestion
> to
> limit the classified traffic but it won't do anything when there is
> congestion becasue the classified traffic can't reach the shaping upper
> limit due to congestion.
>
> "bandwidth" command is only kicked in when there is congestion to
> guarantee
> the availabiloty of the configured bandwidth
>
> Jongsoo
>
> On Wed, 30 Mar 2005 17:55:13 -0500, Sundar Palaniappan <sundarp@gmail.com>
> wrote:
>> Jongsoo,
>>
>> Why would you need to shape and reserve bandwidth for the same traffic
> class?
>>
>> --Sundar Palaniappan
>>
>>
>> On Wed, 30 Mar 2005 12:31:49 -0500, Jongsoo kim <bstrt2002@gmail.com>
> wrote:
>> > OK
>> >
>> > This is my understanding of your emial below.
>> >
>> > When there is no congestion,
>> > class2 can't use more than 200k becasue be = 0
>> > class3 can use up to 400K as a burst with average of 200K becasue bc
>> > = be ( by default )
>> >
>> > When there is congestion, are all class1~3 gauranteed to use 100Kbps
>> > due to Bandwidth 100 command???
>> >
>> > Thanks
>> >
>> >
>> > Jongsoo
>> >
>> > On Wed, 30 Mar 2005 00:52:46 -0500, Sundar Palaniappan
>> > <sundarp@gmail.com> wrote:
>> > > Jongsoo,
>> > >
>> > > Here's my take on this.
>> > >
>> > > Actually, class2 can't use more than 200k. Reason being your
>> > > target and average rate is 200k. There's no burst allowed here.
>> > >
>> > > Class3 can peak up to 400k (200k average+200k burst). You can use
>> > > up to 400k if there is accumalated tokens. You are guaranteed a
>> > > minimum of 200k (average).
>> > >
>> > > HTH,
>> > > Sundar Palaniappan
>> > >
>> > >
>> > > On Tue, 29 Mar 2005 23:31:45 -0500, Jongsoo kim
>> > > <bstrt2002@gmail.com>
> wrote:
>> > > > Guys
>> > > >
>> > > > Let's I have the following policy-map
>> > > >
>> > > > Policy-map shaping
>> > > > Class class1
>> > > > bandwidth 100
>> > > > Class class2
>> > > > bandwdith 100
>> > > > shape averagre 200000
>> > > > class class3
>> > > > bandwdith 100
>> > > > shape peak 200000
>> > > > I apply it to 1000 Kbps serial link.
>> > > >
>> > > > This is my interpretation. Let's set aside 75% of physical BW
> issue.
>> > > > 1) class 1 will behaves it will use as much BW as I can use
>> > > > But when there is congestion, it is guaranteed to have
>> > > > 100Kbps as a minimum
>> > > > 2) class 2 will be shaped to 200Kbps during non-congestion.
>> > > > But when there is congestion, it is guaranteed to have 100Kbps
>> > > > as a minimum
>> > > >
>> > > > 3) class 3 will use up to 200Kbps during non-congestion.
>> > > > But when there is congestion, it is guaranteed to have 100Kbps
>> > > > as a
> minimum.
>> > > >
>> > > > A little confusion I have is the difference between 2) and 3).
>> > > >
>> > > > I believe class 2 can exceed a liittle bit over 200Kbps but in a
>> > > > long run, it will be shaped to 200Kbps while class 3 can't use
>> > > > more than 200Kbps, if BW reaches 200Kbps, thus any exceeding
>> > > > packets will be "dropped"( ??)
>> > > >
>> > > > Basically shape averagre has burst capability but shape peak
>> > > > doesn't have burst capability.
>> > > >
>> > > > Is this correct understanding?
>> > > >
>> > > > Thanks
>> > > >
>> > > > Jongsoo
>> > > >
>> > > > ________________________________________________________________
>> > > > _______ Subscription information may be found at:
>> > > > http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Tue May 03 2005 - 07:54:52 GMT-3