Re: Doubt in CAR : Rate limit

From: Chris Lewis (chrlewiscsco@gmail.com)
Date: Tue Jun 20 2006 - 14:15:42 ART


Koen,

I have posted on this many times, however I'm happy to continue doing so :)

Burst parameters are in bits or bytes, not a rate, if a question states a
rate, either the person writing the question does not understand this, or it
is a typo. The third (and IMHO dubious) alternative as I stated before is
that some explicit reference to available credit is made and an excess rate
for a specific and short amount of time is to be available.

There is no Tc in policing as there is in shaping. Policing calculations are
performed as a packet arrives rather than at regular intervals of Tc as
shaping does. If packet 1 arrives at time T1 and packet 2 arrives at
time T2,

(T2-T1)*CIR worth of tokens are added to the bucket.

The size of the packet is then removed from the bucket and the detemination
of what to do with the packet (conform exceed violate) is taken.

Chris

On 6/20/06, Koen Zeilstra <koen@koenzeilstra.com> wrote:
>
> Hi,
>
> If the questions states configure a burst of 150 kbytes then I would
> presume to use bc 150000/8. From what I read it says kbps so a rate is
> mentions. Or do you mean the question is worded wrong and it should state
> 150 kb instead of kbps. I hope we don't misunderstand each other.
>
> What show and debug commands do you use to find out Tc for CAR? When you
> use 'show interface rate-limit', it won't give you the Tc.
>
>
> quote from:
>
> http://www.cisco.com/univercd/cc/td/doc/product/software/ios123/123cgcr/qos_r/qos_q1g.htm#wp1080850
>
> [quote]
> Policing Traffic with CAR
>
> CAR embodies a rate-limiting feature for policing traffic. When policing
> traffic with CAR, Cisco recommends the following values for the normal and
> extended burst parameters:
>
> normal burst = configured rate * (1 byte)/(8 bits) * 1.5 seconds
>
> extended burst = 2 * normal burst
>
> With the listed choices for parameters, extensive test results have shown
> CAR to achieve the configured rate. If the burst values are too low, then
> the achieved rate is often much lower than the configured rate.
>
> [/quote]
>
> I would conclude the timing interval is 1.5 seconds.
>
> regards,
>
> Koen
>
>
> -----------------------
> "He's just a politician trying to save both his faces ..."
>
> On Tue, 20 Jun 2006, Godswill Oletu wrote:
>
> | Koen,
> |
> | Too many things can go wrong with your approach. I do not see it
> necessary to
> | divide by 1.5.
> |
> | I believe the confusion is how the Bc and Be values were referenced in
> the
> | question. The metric for Bc & Be is per timing interval and not per
> 'second'
> | or anything base on a rate as indicated in the question.
> |
> | One indication that, your assumption might not be correct is that, you
> used
> | the 1.5 on the Bc & Be and left the CIR value. And if you go back to do
> some
> | debugs or some show commands that will tell you what was actually
> applied to
> | the interface, you will discover that the Tc value used was not 1.5.
> |
> | So, converting the value given from bits to bytes by dividing by 8 is
> enough,
> | leave the 'second' (kpbs) referred to by the question.
> |
> | HTH
> | Godswill Oletu
> |
> |
> | >
> | > ----- Original Message ----- From: "Koen Zeilstra" <
> koen@koenzeilstra.com>
> | > To: "Maximus" <victorius@gmail.com>
> | > Cc: "Cisco certification" <ccielab@groupstudy.com>
> | > Sent: Tuesday, June 20, 2006 7:46 AM
> | > Subject: Re: Doubt in CAR : Rate limit
> | >
> | >
> | > > if the b is lowercaps its bits so you should divide by 8. That
> give's you
> | > > the burst size per timing interval. Not per second. So you have to
> convert
> | > > the bps to a byte per timing interval value. I believe the timing
> | > > interface for CAR is 1,5. So that makes 300000/8*1,5=56250.
> | > >
> | > > -> solution 3.
> | > >
> | > > Don't forget the "exceed-action drop".
> | > >
> | > > HTH
> | > >
> | > > Koen
> | > >
> | > > -----------------------
> | > > Kerr's Three Rules for a Successful College:
> | > > Have plenty of football for the alumni, sex for the students,
> | > > and parking for the faculty.
> | > >
> | > > On Tue, 20 Jun 2006, Maximus wrote:
> | > >
> | > > | ok i am only confused on converting the normal and max burst
> | > > |
> | > > | i will give the three kind of answers which i got from pals... can
> | > somebody
> | > > | tell me which one to go forward with..
> | > > |
> | > > | Qn again : the requiremenet is "traffic is limited to 8 Mbps, with
> a
> | > normal
> | > > | burst size of 150kbps and an Max Burst size of 300kbps. traffic
> that
> | > > | conforms is sent
> | > > | with an IP precedence of 5. Traffic that does not conform is
> dropped. "
> | > > |
> | > > | Sol 1 :rate-limit output 8000000 150000 300000 conform-action
> | > > | set-prec-transmit 5
> | > > |
> | > > | Sol 2 :rate-limit output 8000000 18750 37500 conform-action
> | > > | set-prec-transmit 5
> | > > |
> | > > | 150000/8 = 18750
> | > > | 300000/8 = 37500
> | > > |
> | > > | Sol 3 :rate-limit output 8000000 28125 56250 conform-action
> | > > | set-prec-transmit 5
> | > > |
> | > > | (150000/8)*1.5 = 28125
> | > > | (300000/8)*1.5 = 56250
> | > > |
> | > > |
> | > > | On 6/20/06, Vishal Patel <vpatel@accessproviders.com.au> wrote:
> | > > |
> | > > | > In CAR command , the first figure i.e. 8 Mb in your case is in
> Bits
> | > per
> | > > | seconds.
> | > > |
> | > > | Then the second figure is Normal burst which is number of Bytes.
> | > > |
> | > > | Then the third value is the Maximum Burst which is number of
> Bytes.
> | > > |
> | > > |
> | > > |
> | > > |
> | > > | -----Original Message-----
> | > > | From: nobody@groupstudy.com [mailto: nobody@groupstudy.com] On
> Behalf Of
> | > > | Maximus
> | > > | Sent: Tuesday, 20 June 2006 11:21 AM
> | > > | To: Cisco certification
> | > > | Subject: Fwd: Doubt in CAR : Rate limit
> | > > |
> | > > | sorry made a small typo... this is the exact config i have done
> | > > |
> | > > | hie all just a lil confusion
> | > > |
> | > > | the requiremenet is "traffic is limited to 8 Mbps, with a normal
> burst
> | > size
> | > > | of 150kbps and an Max Burst size of 300kbps. traffic that conforms
> is
> | > sent
> | > > | with an IP precedence of 5. Traffic that does not conform is
> dropped. "
> | > > |
> | > > |
> | > > | rate-limit output 8000000 150000 300000 conform-action
> set-prec-transmit
> | > 5
> | > > |
> | > > | is my config rite ? the confision is in normal n max is it in
> bytes or
> | > > | bits...somebody please lead me out...
> | > > |
> | > > |
> | > > | regards,
> | > > |
> | > > | Max
> | > > |
> | > > |
> | > > |
> | > > | -- | regards,
> | > > |
> | > > | Max
> | > > |
> | > > |
> _______________________________________________________________________
> | > > | 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 : Sat Jul 01 2006 - 07:57:33 ART