The IP over ATM Mailing List Archive by date

[Mark Laubach <laubach@terra.com21.com>]

> If the ATMARP requests are necessary then the InATMARP
> requests appear redundant for SVCs. If they are redundant 
> can we get rid of them (neatly solving the data loss problem
> due to the lack of a 3-way handshake in Q93B :-)

You raise a good point that the InATMARP requests may appear
redundant in the planned rewritten world.  We hashed through
the mechanism a month or so ago.  The result of that 
discussion was to make the ATMARP server operate in more of
a promiscuous manner by requiring the client to do initial
registration via an ATMARP_Request.  The server would merely
sniff out the source information to build it's table. Carrying 
it forward, that means that InATMARP's won't be needed for 
initial registration.  It's good to have them around to use
on open VCs. I wasn't planning to start any rewrites of
the mechanism until the first of the year - about the time
we'll see some interoperability results from various
implementations.

Mark