From: Carlos G Mendioroz (tron@huapi.ba.ar)
Date: Thu Jul 19 2007 - 08:00:55 ART
I disagree here...
there is no way to measure bandwidth w/o a time reference.
So we need to know the Tc for calculating the burst size, but in any
case, the frasing of the question is wrong.
(There is no way to impose a two rate policer with a one bucket algo).
The page says "The default burst size is calculated based on a
200-millisecond interval and the bandwidth configured for low latency
queueing (LLQ)." so we can infer that the Tc is 200ms, but is not
clearly stated.
Bursting allows to go over speed limit because the algo (token bucket)
has no other long term memory, and is the bucket size in bytes.
LLQ should be CBR like, so the size should not matter that much,
but if you use it for VBR, then ...you need this.
-Carlos
Mike Kraus (mikraus) @ 4/7/2007 13:25 GMT -0300 dixit:
> Antonio,
>
> You are correct, but I'm going to attempt to clarify a bit, as I feel
> that statement might confuse the issue to some. The documentation is
> stating that the default burst size (Bc) is calculated off of a 200ms
> interval of the configured mean rate (CIR), not that there is an
> interval of 200ms that is somehow actually used or calculated by the
> device.
>
> The concept behind LLQ was to ensure voice was passed as quickly as
> possible, so the priority queue is a policer. So, there is no
> buffering, hence no Tc. They had to use some formula to calculate Bc
> off of the rate, or they could have just assumed a default value (like
> 1500 bytes in policing in MQC).
>
> Thanks,
> Mike
>
>
> -----Original Message-----
> From: Antonio Soares [mailto:amsoares@netcabo.pt]
> Sent: Wednesday, July 04, 2007 11:29 AM
> To: 'BitGossip'; Mike Kraus (mikraus); ccielab@groupstudy.com
> Subject: RE: MQC burst size
>
> The feature document states that the interval is 200 ms:
>
> http://www.cisco.com/univercd/cc/td/doc/product/software/ios121/121newft
> /121
> t/121t3/dtcfgbst.htm
>
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> BitGossip
> Sent: quarta-feira, 4 de Julho de 2007 16:45
> To: 'Mike Kraus (mikraus)'; ccielab@groupstudy.com
> Subject: RE: MQC burst size
>
> Mike,
> The requirement asks for burst of 64kbps while the command expects a
> figure in byte To first number is dimensionally bit per second while the
> second is byte or bit.
> So to go from the first to the second we have to multiply by a time
> interval
>
> Burst size (byte) = burst (in bps) * time_interval / 8 (byte per bit)
>
> So what is this time_interval ?
>
> This is very similar to BC=CIR * Tc
>
> Or I am completely wrong.....
>
> Thanks,
> Bit.
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> Mike Kraus (mikraus)
> Sent: 04 July 2007 14:54
> To: Bit Gossip; ccielab@groupstudy.com
> Subject: RE: MQC burst size
>
> Wouldn't it just be: priority 128 8000 (burst is in bytes)
>
> Perhaps I'm misunderstanding something?
>
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> Bit Gossip
> Sent: Wednesday, July 04, 2007 3:15 AM
> To: ccielab@groupstudy.com
> Subject: MQC burst size
>
> Does anyone knows how to calculate the burst size for the MQC priority
> command?
>
> priority {bandwidth-kbps | percent percentage} [burst]
>
> For instance if the maximum rate is 128kbps and I want to allow burst of
> 64kbps
>
> _______________________________________________________________________
> 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
>
-- Carlos G Mendioroz <tron@huapi.ba.ar> LW7 EQI Argentina
This archive was generated by hypermail 2.1.4 : Sat Aug 18 2007 - 08:17:41 ART