The IP over ATM Mailing List Archive by date

[Curtis Villamizar <curtis@ans.net>]

In message <199507181727.CAA15960@necom830.cc.titech.ac.jp>, Masataka Ohta writ
es:
> > >Shortcut paths on a large scale isn't going to work.
> > 
> > Not if done blindly, I agree.  That's why I propose an architecture that
> > limits fan-in/fan-out by using intermediate IPmc routers within large
> > ATM clouds.  I don't think any other solution will fix scaling problems
> > with large IP multicast sessions running over large ATM clouds.
> 
> Putting intermediate IP routers within large clound is, by definition,
> make the cloud a collection of smaller clouds.
> 
> So, you are not running multicast session over large could.
> 
> The other problem is that it is not your proposal but mine.
> 
> 						Masataka Ohta
> 


I think there is a mjor difference in the two proposals.  Please
correct me if I am wrong since I don't read the colip list, but
doesn't COLIP propose a cell by cell forwarding?  This is more of an
IP/ATM hybrid approach.  I think Walter is suggesting more of a pure
IP approach, breaking the network into large LIS.  As cells arrive
they are reassembled into frames and making an IP level decision and
then forwarding to another LIS.  The use of MARS allows the LIS to
grow larger, reducing the number of multicast routers needed.

Did I understand these proposals correctly?

Curtis