From: Carlos G Mendioroz (tron@xxxxxxxxxxx)
Date: Tue Jun 25 2002 - 08:48:48 GMT-3
Sorry for late reply, but...
Even though it has been said many times, I think it's wrong to put CIR =
AR
in many cases.
CIR is your "want to be transmiting at this speed" goal. And it's not
a hard limit on your tx speed. AR is.
You can transmit above CIR speed if you have some tokens to spare.
(I.e. you can transmit up to Be bits above CIR, you can not sustain that
though).
minCIR is a cisco specific thing that is related to shaping, and is the
minimum
speed you will agree to transmit when forced by congestion indication.
(BECN)
Many of the settings only make sense in contexts. If you have a provider
that is
policing you, and will send your traffic to /dev/null if you run above
your
excess rate, there is no good in setting CIR = AR.
On the other hand, if the network is yours, then why would you waste the
BW when
nobody else is using it ? It all depends.
I found http://www.cisco.com/warp/public/125/traffic_shaping_6151.html
to be very
clear about it.
MADMAN wrote:
>
> Anthony Pace wrote:
> >
> > But if each map-class has a CIR and MINCIR, don't I want each one to
> > have only a peice of the pie? If I read you correctly you would want
> > all 8 map-classes to have a CIR of 1544 so that they "may" use all the
> > bandwidth if there were no congestion and a MINCIR of say 128 so that
> > if every spoke had traffic we would diveid it up. Is that correct?
>
> In a nutshell, yes. Otherwise your basically screwing yourself out of
> potential bandwidth. If you happen to be going in and out of the same
> frame switch, no NNI, you can often get well above your CIR for a
> sustained time.
>
> > Also does the command wjere traffic shaping parms are all on one line
> > express the Bc abd Be different than if you put thme seperatly in a
> > map-class?
>
> Not sure what you mean.
>
> Dave
> >
> > Anthony Pace
> >
> > On Thu, 20 Jun 2002 13:45:53 -0500, "MADMAN" <dave@interprise.com>
> > said:
> > >
> > >
> > > Anthony Pace wrote:
> > > >
> > > > I also keep seeing this reference to setting the CIR to the port speed
> > > > in all the CCO an in all the books and in these posts. I guess I will
> > > > adhere to this in the lab since I am seeing that this CCIE effort is
> > > > more theory than practical, however, what I have observed in real life
> > > > is that traffic shaping is used when many, many logical circuits
> > > > inhabit a phisical link (ATM or FRAME). In the case of frame-relay, if
> > > > the hub has 1544 kbps and 8 pvc's with 8 class's and each spoke has 64k
> > > > on each of their circuits,how can the CIR be 1544 on every map class.
> > > > Isn't the idea to carve up the bandwidth amongst the spokes?
> > > >
> > > If you have a 1.544M circuit and you configure FRTS CIR for say 128K
> > > you will have effectively limited yourself to a max of 128K and will
> > > not
> > > be able to burst at a higher rate. Yes you want to shape when needed
> > > but you also want to burst when the bandwidth is available.
> > >
> > > Dave
> > >
> > > > Anthony Pace
> > > >
> > > > On Thu, 20 Jun 2002 22:31:25 +0800 (HKT), "Katson Yeung"
> > > > <kyeung@mail.hkcix.com> said:
> > > > > Thanks Mamoor, that is correct.
> > > > >
> > > > > One more question about the CIR.
> > > > >
> > > > > According to all the materials I read, the recommended value for CIR
is
> > > > > set it to Access-rate. So, there is a question:
> > > > >
> > > > > R1s0 ------ FR -------- s0R2
> > > > >
> > > > > R1s0 = 1.544Mbps
> > > > > R2s0 = 128kbps
> > > > > CIR = 64k. (I know it is = mincir, irrelevent to the question, but li
st
> > > > > it
> > > > > out anyway.)
> > > > >
> > > > >
> > > > > At R1,
> > > > > should I set the CIR as 1544000 or 128000?
> > > David Madland
> > > Sr. Network Engineer
> > > CCIE# 2016
> > > Qwest Communications Int. Inc.
> > > dave@interprise.com
> > > 612-664-3367
> > >
> > > "Emotion should reflect reason not guide it"
> > >
> >
> > --
> > Anthony Pace
> > anthonypace@fastmail.fm
> >
> > --
> > http://fastmail.fm
> > - Ever wonder why we aren't named snailmail.sm?
>
> --
> David Madland
> Sr. Network Engineer
> CCIE# 2016
> Qwest Communications Int. Inc.
> dave@interprise.com
> 612-664-3367
>
> "Emotion should reflect reason not guide it"
This archive was generated by hypermail 2.1.4 : Tue Jul 02 2002 - 08:12:41 GMT-3