The IP Over NBMA (ION) Archive

Cell Relay Retreat>ION Archive>month:1996-Jul> msg00073



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

ATMARP server source code/"classic IP"

  • From: Albert Manfredi <manfredi@engr05.comsys.rockwell.com>
  • Date: Fri, 12 Jul 1996 17:51:16 -0700
  • CC: ion@nexen.com
  • Organization: Rockwell Defense Electronics - Collins

brutzman@cs.nps.navy.mil wrote:

> I am curious about the genesis of the term "classic IP."  Can anyone tell
> me where this came from.

I think it simply refers to the standard structure of Internet: 
individual, small LANs with a flat address space used to identify 
interfaces locally, and use of Layer 3 (IP) addressing, routing, and 
routers to interconnect these LANs.

This is not what ATM was designed to operate under.

> In particular it appears to be a euphemism for
> unicast IP, i.e. IP minus many-to-many multicast.

I don't think so, except that Internet was (and is still) far more 
unicast than multicast oriented. I guess unicast examples are more 
natural.

> The term "end-to-end" above appears to assume point-to-point connections
> to match the connection-oriented nature of ATM.

Actually, in my opinion, this is certainly _not_ the case. If anything, 
the fact that ATM sets up circuits (I really think it should be described 
as "circuit switched" in this context, rather than connection oriented) 
becomes an absolute nuisance in RFC-1577, which looks to use ATM as an 
Ethernet replacement. It's not that ATM circuits might not somehow be 
matched up with TCP/IP connections, conceptually. Rather, the nuisance 
factor results from the fact that this is not how Ethernet is used by IP, 
and therefore RFC-1577 has to make these ATM circuits emulate a broadcast 
medium. I'm saying that they specifically can't allow TCP/IP connections 
to "match the connection oriented nature of ATM."

Bert
manfredi@engr05.comsys.rockwell.com