From: "Norman W. Finn" <nfinn@cisco.com>
Subject: Re: ["Robin Tasker": Priority in 802.1p]
To: crawdad@FNAL.GOV (Matt Crawford)
Date: Thu, 11 Apr 1996 13:03:56 -0700 (PDT)
Cc: nfinn@cisco.com, P8021@nic.hep.net


Matt Crawford writes:
> 
> > ... Also, there is only one group
> > address per flow, with fewer flows than endstations.  ...
> 
> How do you figure?  I can imagine 2^N - 1 flows, each with a
> different non-empty set of receivers among N endstations.

You've got a longer vision than I, Matt.  :)  I guess we can imagine
that.  But the control packets to support that scenario are still cheaper
than broadcasting all that multicast data everywhere.

Unless ...

One thing we haven't talked about is the cost of maintaining a MAC-layer
multicast tree in the absence of data flowing on that tree.  I would
hope that the upper layers would take the trouble to kill unneeded,
inactive, multicast groups.

-- Norm