Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

[RPRWG] default ringlet selection





In the current version of the draft, P802.17/D2.2,
the requirement for default ringlet selection is stated
as follows:

"The algorithm used to choose ringlets is implementation
specific.  The only constraint on the algorithm is that
it chooses the same direction for all frames with the
same {da, sa, sc} tuple, until protection or topology
changes."

I find the above requirement overly restrictive for
someone wanting to provide better algorithms for 
ringlet selection.  For example, link aggregation in
802.3 imposes the ordering requirement for conversations
which is at a much finer granularity than what
the MAC header provides.  A conversation is defined
as follows:

"Conversation: A set of MAC frames transmitted from one 
end station to another, where all of the MAC frames form 
an ordered sequence, and where the communicating end 
stations require the ordering to be maintained among 
the set of MAC frames exchanged. (See IEEE 802.3 Clause 43.)"

I had submitted a comment about this on P802.17/D2.1 
(Comment #309), proposing that we adopt a similar 
requirement for default ringlet selection.  It was rejected 
with the explanation that it "over-complicates the MAC 
architecture".  I don't understand how allowing a variety 
of behaviors that has no impact whatsoever on other MACs 
in the ring leads to over-complication in the architecture.  
Can someone that diagreed with the comment at the meeting 
please explain?

-Anoop
--
Anoop Ghanwani - Lantern Communications - 408-521-6707