Hello folks,
Here is the main body of the text of the RFC 6621 considerations
document. I expect to have it ready today, depending on various
turkey-related factors. Comments will be appreciated. It's missing all
the boilerplate and citations, but I'll get that done soon.
------------------------------------
Limiting the number of multicast forwarders provides performance benefits
when used with a protocol that relies on flooding messages throughout a
network. Methods for limiting the number of multicast forwarders while
maintaining complete network coverage are specified in RFC 6621, and AODVv2
can offer much greater performance using RFC 6621 algorithms. These
algorithms
do not affect the origination of multicast messages, but only restrict
whether
or not a node should forward incoming multicast messages that are
intended to
flood the network. In fact, enabling this performance benefit for reactive
manet protocols was a strong motivating factor in the development and
publication of RFC 6621.
RFC 6621 algorithms are specified to operate at layer 3. AODVv2 uses
RFC 5444
as a message bundling layer. Moreover, AODVv2 specifies that multicast
control
messages have to be processed before regeneration; in other words, the
incoming
multicast message is not forwarded at all. This document specifies a
way to use
RFC 6621 algorithms at the layer above RFC 5444 so that AODVv2 can reap the
benefits of the multicast suppression algorithms in RFC 6621.
For this purpose, define a multicast forwarding node as a node that
currently forwards incoming multicast messages to its neighbors, based
on the results of running a multicast suppression algorithm as specified
in RFC 6621.
If the node is not currently a multicast forwarding node, then the
node does not regenerate multicast messages (e.g., RREQ or RERR) even
if the conditions for regeneration (cite AODVv2) are satisfied. This
restriction
only applies for regeneration, and does not apply for multicast messages
originated by the node.
-----------------------------------
Regards,
Charlie P.
