The IP over ATM Mailing List Archive by date

Cell Relay Retreat>List Archive>month:1995-Jan> msg00102



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

IP Architecture Extensions for ATM

  • From: hiroshi@ctr.columbia.edu (Hiroshi Esaki)
  • Date: Tue, 31 Jan 1995 15:30:25 -0500
  • CC: yakov@watspn.ibm.com, ip-atm@hplb.hpl.hp.com


Hi, 

  >> The alternative that this document proposes is odd: we use IP routers,
  >> but when applications need a direct SVC, they subvert the routing:
  >> There are two approaches to this
  >>
  >> A/ A router helps - bascally the scenario is
  >>
  >>	 LIS 1                           LIS 2
  >> ^^^^^^^^^^^^^^^^^^              ^^^^^^^^^^^^^^^^^
  >>		       .....router...
  >>	 	       :            :
  >> src ..switch.... switch ..... switch..switch ....dst
  >>		      x            y
  >>
  >> The src simply sends IP packets "to" the router. The router,  after
  >> the initial packet (which could have been the ARP packet) tells its
  >> immediate switches to hand off the vci/vpi from x to it, and it to x
  >> and y to it and it to y, to go thru the ports between the switches

Just a comment. 
When switch with x, switch with y and a router is in a one
(we called this as a router in our internet-draft), it is exactly 
what Ohta-san and I was proposed in draft-esaki-co-cl-forw-atm-00.txt. 
We proposed VCI/VPI value for a routing cache of IP packet.

Thanks, 

Hiroshi Esaki