The IP over ATM Mailing List Archive by date

[Sridhar Komandur <komandur@cs.pitt.edu>]

   >   From: taddy@cc.gatech.edu (Rajesh Talpade)
   >   Date: Thu, 28 Mar 1996 11:12:37 -0500 (EST)
   >
   >   Yes, MCS would still be needed. 
   >
   >   Having the mp-to-p capability would mean all senders would open one
   >   mp-to-p connection to each receiver. This is similar to a VC Mesh,
   >   except that it is inverted, in that each mp-to-p VC is rooted at a
   >   receiver, ie the 'mp' in the mp-to-p are the senders, while the 'p'
   >   is each receiver.


Hi,

If mp-to-p capability is available, with no reassembly at the intermediate
switches - I guess this means cell interleaving is possible, would it be 
necessary for each sender to open a mp-to-p connection to each receiver? 
I wonder why not go an extra step and support mpt-mpt if the following is 
true in general (+ assuming the number of *concurrent* senders is small)? 

   >	<deleted>
   >   significantly more receivers than senders in a group, and hence more
   >   mp-to-p VCs.

Regards,
Sridhar Komandur
komandur@cs.pitt.edu