From: Ted Richmond (rich_ted@xxxxxxxxx)
Date: Tue Aug 20 2002 - 15:18:01 GMT-3
Thanks Edward, Michael and Elping for your inputs.
I still am not clear on Be. Can it be less that Bc?
Secondly, what do you guys think about the example in
the doc where CIR - 1 bit/sec? IS it still valid?
Ed, I guess you understood that example, can you
please clarify?
TIA again!!!
--- Edward Monk <emonk@att.net> wrote:
> Ted,
>
>
> This is an interesting link on how to set your Be
> higher than your Bc.
> This allows you to save up enough tokens for future
> bursts that exceed
> your token savings during certain intervals. This
> only works when you
> have enough excess bandwidth available. Which should
> work in most cases.
>
> Excerpt from document below: This following
> paragraph is a key excerpt
> from the document and should help you to
> understanding what they are
> doing here and how they achieve the desired result.
>
> The bucket itself has a specified capacity. If the
> bucket fills to
> capacity, newly arriving tokens are discarded and
> are not available to
> future packets. Thus, at any time, the largest burst
> a source can send
> into the network is roughly proportional to the size
> of the bucket. A
> token bucket permits burstiness, but bounds it.
>
> Configuring the functionality of the token bucket
> with a Be greater than
> Bc in this manner helps to avoid what is called tail
> loss. Tail loss is
> when a sudden dropping of packets occurs because you
> have run out of
> tokens saved up in the token bucket. This
> configuration actually allows
> for a more gradual dropping of packets while the
> bandwidth is being
> throttled back. It actually prevents Bc from
> depleting the Be (token
> bucket) to rapidly because Be is a larger value than
> Bc.
>
> Very interesting thanks for the link. Another good
> day learned something
> new.
>
> -----Original Message-----
> From: nobody@groupstudy.com
> [mailto:nobody@groupstudy.com] On Behalf Of
> Ted Richmond
> Sent: Tuesday, August 20, 2002 2:15 AM
> To: ccielab@groupstudy.com
> Subject: Bc Be S.O.S.....
>
> Hello group,
> When I finally thought that I've got Qos
> under-control
> (err...well say 80% undercontrol), I came across a
> doc
> - Selecting Burst and Extended Burst Values for
> Class-Based Policing
>
>
http://www.cisco.com/warp/public/105/carburstvalues.html
>
> This one really threw me off. Here are some
> questions:
>
> * Are Be,Bc definitions different for shaping and
> policing?
>
> * The above doc says, "When bc is equal to be, the
> traffic regulator cannot borrow tokens and simply
> drops the packet when insufficient tokens are
> available". I though this is the case when be=0
>
> * The doc gives an example with 'CIR= 1 bit/sec, Bc=
> 2
> bytes/sec and Be=4 bytes/sec' (pkt size as 1 byte).
> Is
> this possible? How can he send 4 packets with the
> above constraints and still maintain 1bit/sec CIR?
>
> I am totally lost now - I should've never read this
> doc. Please help -S.O.S
>
> TIA.
>
This archive was generated by hypermail 2.1.4 : Sat Sep 07 2002 - 19:48:30 GMT-3