The IP over ATM Mailing List Archive by date

Cell Relay Retreat>List Archive>month:1996-Mar> msg00213



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

EE Times on IP over ATM

  • From: Grenville Armitage <gja@bellcore.com>
  • Date: Thu, 28 Mar 1996 11:14:15 -0500
  • cc: gja@bellcore.com, ip-atm@nexen.com

>>my question to you is, if atm
>>forum one day would define aal_sdu aware mp-to-p atm connections in
>>addition to the current p-to-p and p-to-mp connections, would mcs still
>>be needed?

An interesting question. Most commonly an MCS is used to emulate
mp-mp service, in effect it concatenates an emulated mp-p with a p-mp
to do so. My gut feel is that if mp-p calls became possible through
appropriate switch support, then a virtual MCS could be achieved
by explicitly piping the output of the mp-p into the top of
a p-mp VC.

This may not entirely eliminate the need for a special point in
the data path where the mp-p is piped into the p-mp. Currently
the MCS terminates 1 VC per sender. If the MCS were being fed
by a network supported mp-p call, then the MCS itself would
reduce down to a node that terminates a single VC (the incoming
mp-p) and originate a single VC (the outbound p-mp).

cheers,
gja