Re: [Diffserv] Comments on the TCB of the Conceptual Model - msg 1 of 2

["John Strassner" <[e-mail]>]

Comments inline.

regards,
John
----- Original Message -----
From: "Michael Richardson" <[e-mail]>
To: <[e-mail]>
Sent: Thursday, July 06, 2000 8:31 AM
Subject: Re: [Diffserv] Comments on the TCB of the
Conceptual Model - msg 1 of 2


>
> >>>>> "John" == John Strassner <[e-mail]>
writes:
>     John> classification, metering, actions, and queuing.
What first
>     John> confused me was the "action" category.
Classifiers, meters,
>     John> markers, droppers, and queues all are "actions"
that are
>     John> taken on the packet (whereas, for example,
counters are not,
>     John> even though they are included in your action
category). So I
>     John> was trying to understand how you defined an
"action".
>
>   I don't agree at all.
>   Queueing and marking might be actions to be taken on the
packet.
>
>   But, classifing, and metering are things that the system
does with
> the packet to establish properties about the packet and/or
flow.

<js>
Au contraire, each of these five functions can be modeled as
a black box that has one or more inputs and one or more
outputs that performs an action on the traffic. So to take
one simple counter-example, a classifier consists of a set
of filters that are used to separate a single input stream
into multiple output streams. Thus, I disagree with you on
two points:

  1) the classifier does not establish properties about the
packet/flow,
     rather it uses filters to split the flow into separate
flows
  2) the action of filtering is clearly an action

The same argument can be made for metering and dropping.

Therefore, from a modeling point-of-view, one can view these
five functions as a set of building blocks that can be
arranged as appropriate (including having multiple copies of
the same building block) to define how particular
conditioning is performed on a given flow. Which in turn
means that they have something in common (as another answer
for the question below).

>     John> approach, I would have preferred to see
classifiers, meters,
>     John> markers, droppers, and queues all as subclasses
of a more
>     John> general class that was used as the basis of
conditioning
>     John> traffic. The model as currently described
implies that these
>     John> are very different things with not a lot in
common.
>
>   I don't think that they have anything in common.

Of course they do. They all perform actions on the packet.
They all interconnect with each other. They all work
together to condition one flow in a manner that is different
for the conditioning of other flows. etc.

>    :!mcr!:            |  Solidum Systems Corporation,
http://www.solidum.com
>    Michael Richardson |      ... visiting customers in San
Jose.
>  Personal: <A
HREF="http://www.sandelman.ottawa.on.ca/People/Michael_Richa
rdson/Bio.html">[e-mail]</A>. PGP key
available.
>  Corporate: <A
HREF="mailto:[e-mail]">[e-mail]</A>.
>
>
>
> _______________________________________________
> diffserv mailing list
> [e-mail]
> http://www1.ietf.org/mailman/listinfo/diffserv
> Archive: http://www-nrg.ee.lbl.gov/diff-serv-arch/
>
>


_______________________________________________
diffserv mailing list
[e-mail]
http://www1.ietf.org/mailman/listinfo/diffserv
Archive: http://www-nrg.ee.lbl.gov/diff-serv-arch/