From: Kumar Raja-Q16843 (Raja.Kumar@motorola.com)
Date: Wed Nov 16 2005 - 11:13:16 GMT-3
Crystal Clear .... thanks a Million Chris ..... Raja
________________________________
From: Chris Lewis [mailto:chrlewiscsco@yahoo.com]
Sent: Wednesday, November 16, 2005 7:34 PM
To: Imal kalutotage; Kumar Raja-Q16843
Cc: ccielab@groupstudy.com; Daniel Berlinski
Subject: Re: Question on Qos rate-limiting
I've posted on this topic a few times, and here is a link to one of the
longer responses
http://www.groupstudy.com/archives/ccielab/200509/msg00978.html
<http://www.groupstudy.com/archives/ccielab/200509/msg00978.html>
I think though you need to split this in to two. The first is what you
need to think of when answering lab questions. The second is how the
algorithms work. I don't believe it is necessary to fully understand the
issues of actual and compound debt as they relate to rate limiting to be
able to correctly answer the lab questions.
The main issue I see with the discssion on this thread so far is that of
the concept of Tc for rate limiting. Tc in terms of a regular interval
at which bits are transmitted or at which time tokens are added to the
bucket is a concept from shaping, and has no relevance to the rate !
limit (police) algorithms. THis is discussed at the link above.
The other main point of confusion for this topic seems to be thinking of
burst parameters in terms of a rate, when it is in fact a measurement of
bytes, not the rate of transmission of those bytes.
My feeling is this. If you are given burst parameters in a question that
are specified in a bits vlaue, divide it by 8 to get the bytes. If no
burst is specified, leave it at default.
Chris
Imal kalutotage <imal.kalutotage@gmail.com> wrote:
Hi Kumar,
What I remember is multiply by 1.5 sec is not needed..
May be Chris can explain this very well..
Imal
On 11/16/05, Kumar Raja-Q16843 wrote:
>
> Thanks Daniel for your inputs, let me r! ephrase the question,
just want
> to be doubly sure, this was a wicked question thrown at us
during the
> practice, but then it make sense......this question makes me
remember
> the response of Tim in the mail train "Top Reasons People Fail
the Lab",
> Coming to the point...
>
> CIR = 10 Mbps; Normal Burst = 500 kbps; Maximum Burst = 1 Mbps
>
> rate-limit input
> Bytes> transmit exceed-action drop
>
> The attribute CIR is straight forwards since it is in
bits/sec, the
> given value is the same measure
> My test is here with the Normal Burst, the attribute in the
command has
> a measure of bytes, let me try (pardon me if I am being
stupid)
> (500*1000/8)*1.5 = 93750 bytes
> Same the case with Maximum burst, the value would be
(1000000/8)*1.5 =
> 187,500 bytes
> Then the command would be
> rate-limit input 10000000 9375! 0 187500 transmit
exceed-action drop
>
> Has my way calculating the values are correct or am I missing
something
> here?
> I guess, , Does it still sound stupid?
>
> Thanks in advance,
> Raja
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On
Behalf Of
> Daniel Berlinski
> Sent: Wednesday, November 16, 2005 9:49 AM
> To: Kumar Raja-Q16843; ccielab@groupstudy.com
> Subject: RE: Question on Qos rate-limiting
>
> Kumar
>
> IMHO your numbers make a Tc of 50ms (bc/cir *1000) hence you
area
> leaking bytes into the line across 20 intervals in one sec to
send 10Meg
> If you allow a burst of 125000 bytes, it means that after 20
intervals
> of 50ms each you will end up sending 125000 bytes(1Mbps) * 20=
20Mbps
>
> To allow a max burst of 1Mbps I think (I could be wrong please
someone
> correct me in this one! ) you should use a be of 6250 bytes
>
> Cheers
>
>
> ________________________________
>
> From: nobody@groupstudy.com on behalf of Kumar Raja-Q16843
> Sent: Wed 16/11/2005 7:11 a.m.
> To: ccielab@groupstudy.com
> Subject: Question on Qos rate-limiting
>
>
>
> Hi Guys,
>
> In rate-limiting , I have this question, a CIR of 10 Mbps with
a normal
> burst of 500 kbps and a maximum burst of 1 Mbps, the traffic
should pass
> if the set condition matches if not the packet be dropped.
>
> Interface Ethernet0/0
>
> ip address 10.10.10.1
255.255.255.0
>
> rate-limit input 10000000 62500 125000 conform-action transmit
> exceed-action drop
>
> Kindly advice if my configuration is correct, the above
solution is
> because I was given the normal burst value of 500 kbps, but
actually we> don't need this value, if we are given with the CIR and the
normal burst
> can be calculated as (CIR*1.5/8), when the value of normal
burst is
> explicitly given, should we use or we still go ahead and
calculate the
> normal using the formula.
>
> Thanks in advance,
>
> Raja
>
>
This archive was generated by hypermail 2.1.4 : Thu Dec 01 2005 - 09:12:07 GMT-3