The MPLS WG Archive

[Eric Gray <ewgray@GraIyMage.com>]

Andreas,

    As I understand it, you are close but not exactly right.

    In the simplest terms, the thing that makes it so that a group of 
(two or more)
Behavior Aggregates cannot be part of the same Ordered Aggregate is when we
either cannot define, or cannot agree on a definition, for a 
deterministic decision
process for handling a near simultaneous arrival of packets from two or 
more BA
within the same (hypothetical) OA - for every possible combination.

    There are two things to note here. The first is that this difficulty 
should have no
effect on ordering within a BA unless the decision making rules are 
pathological.
The second is that this does not mean that clever people have not 
already thought
of ways to make such a decision.

--
Eric

Andreas Hess wrote:

>Hi,
>I am not sure if I really understood the concept of OAs and BAs.
>To my understanding a DSCP value consists of a PHB scheduling class part and a 
>drop precedence part. Furthermore, a BA is identified via the DSCP value.
>So far, things are fine.
>But I am not sure about  the definition of an ordered aggregate (OA), which is 
>said to be a set of BAs that share an ordering constraint. Does this mean 
>that an OA specifies the interaction of multiple PHBs on a diffserv node such 
>that no reordering --- on microflow basis --- of packets occurs?
>Is this correct?
>Thanks for helping!
>
>Regards Andreas
>
>
>
>
>
>
>_______________________________________________
>mpls mailing list
>mpls@lists.ietf.org
>https://www1.ietf.org/mailman/listinfo/mpls
>
>
>
>  
>


_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls