Cell Relay Archive

[albert.e.manfredi@boeing.com]

In article
<Pine.WNT.3.96.980819154054.-208601A-100000@hamming.esat.kuleuven.ac.be>,
  Michel Teughels <Michel.Teughels@esat.kuleuven.ac.be> wrote:
> Hi,
>
> In Q.2971 (about signalling for point-to-multipoint connections)
> they say a network has to reject a SETUP message if the Broadband
> bearer capability IE indicates "point-to-multipoint" and
> contains an ATM traffic descriptor IE with any backward cell rate
> field specifying a non-zero value.
>
> Now why is this ?  Isn't the problem of sorting out which cells
> were sent by which leaf a problem of the higher layers ?

Higher layers? The problem is that we're talking about cells here. Higher
layers cannot differentiate between cells in a shared VC unless you have merge
points packetize the different cell streams from the leaves. It's this merging
that causes the problems for mpt-mpt comms over ATM. Packetizing the cells, of
course, sort of defeats the whole purpose of ATM. So the "right" way to do it
is to establish different return VCs for each leaf. Or to do something that
emulates IP multicast.

> I would like to use a bi-directional pt-to-mpt ATM connection
> between two end-stations (one of which with multiple ATM NICs),
> and use the redundant branches as back-up connections.

Interesting. I suppose that in such a specialized scenario, where the station
with multiple NICs knows never to transmit cells simultaneously on that VC
from both NICs, such a scheme could be cobbled up. But the general case
cannot use such a solution.

Bert
manfredi@arl.bna.boeing.com

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/rg_mkgrp.xp   Create Your Own Free Member Forum