The IP over ATM Mailing List Archive by date

Cell Relay Retreat>List Archive>month:1995-Oct> msg00058



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

Multicasting with ATM - a concern

  • From: Dan Grossman <dan@dma.isg.mot.com>
  • Date: Wed, 18 Oct 1995 13:13:59 -0400

>    There is an alternative that allows AAL5 to be used for mpt-to-mpt
>    connections.  Namely, set up the connection as a switched VP and
>    either: 1) allocate a unique VCI to each transmitter
> 
> this is worth considering now that we have switched VP capability in uni
> 4.0 signalling and thatfwhich many switches have already implemented for
> soft pvp purposes.
> 
> i don't remember how AAL3/4 implements MID allocation, but vci allocation
> protocol could be something similar.
> 
> -- juha
This is highly inadvisable.

Switched VPCs are unlikely to be widely deployed. This is because they make 
inefficient use of a very scarce resource, the routing (VPI + VCI) field of 
the ATM cell header.   VPCs were intended for, and are still needed for, 
aggregating of VCCs for purposes such as traffic management, trunking between 
switches in public and private networks, and failure recovery.  Recently, 
there also have been proposals to also use the VPI as a MAC address in shared 
media residential broadband networks.   VPCs were never intended to be a way 
for applications to get extra header bits, and widespread misuse for this 
purpose is inconsistent with their use for their intended purpose.   

It is true that the signalling group in the ATM Forum did vote by a narrow 
majority to put switched VPCs in their document, then voted by another narrow 
majority to take them out, then voted by yet another narrow majority to put 
them back in.  ITU-T has a fairly strong position that switched VPCs are not a 
service to be offered by public networks.  This has to be an indication that 
industry support for switched VPCs is weak. 

Regardless of whether people share the concern of many of us for the proper 
use of the VPI, serious consideration should be given to whether you really 
want to write an Internet standard assuming the availablity of mechanism which 
is not going to be widely available.

If we're looking for future ways of supporting multipoint-multipoint services, 
consideration should be given to the ongoing work in ITU on calls 
(associations amongst ATM endpoints) with multiple (ATM layer) connections.