The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2001-Jan> msg00054



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

Some queries

  • From: "Cheng-Yin Lee" <leecy@nortelnetworks.com>
  • Date: Fri, 05 Jan 2001 12:05:33 -0500
  • CC: Juha Heinanen <jh@lohi.eng.telia.fi>, "Bagasrawala, Abbas" <abagasrawala@lucent.com>, curtis <curtis@avici.com>, "Faisal S. Naik" <faisal@hamdard.net.pk>, mpls uunet <mpls@UU.NET>, zaziz <zaziz@cisco.com>, prasanna <prasanna@csa.iisc.ernet.in>
  • Organization: Nortel Networks
  • X-Orig: <leecy@nortelnetworks.com>

Yi,
The related point that I was trying to get clarification from Curtis was that it is not so much independent control planes that result in IGP scaling issues (wrt the quote below from Curtis), but rather how fully-meshed one's network is.  But I think this point is clearer now.

thanks,
cyl 

Curtis Villamizar wrote:
There were a number of motivations for getting rid of ATM in IP
networks.  One was the cell tax which at a typical 20% would not alone
kill ATM.  Another was SAR.  You can't get fast ATM router interfaces
because SAR speed and SAR buffering becomes a problem.  This alone was
enough to kill ATM.
Perhaps even larger was the problems of independent control planes and
the effect on IGP scaling. ....

Yi Chu wrote:

So based on the reply from Juha and Abbas, there is no need to have 999
adjacencies for IP over ATM either.  A densely meshed network would work for
ATM solution as well (10 to 15 adjacencies per router for a 1000 router
network).

I understand the need to improve the overlay model, and thus the need for
MPLS.  But I just can not stand false statements.

Yi

-----Original Message-----
From:   owner-mpls@UU.NET [mailto:owner-mpls@UU.NET] On Behalf Of Juha
Heinanen
Sent:   Friday, January 05, 2001 10:11 AM
To:     Yi Chu
Cc:     Bagasrawala, Abbas; curtis@avici.com; Cheng-Yin Lee; Faisal S. Naik;
mpls uunet; zaziz@cisco.com; prasanna@csa.iisc.ernet.in
Subject:        RE: Some queries

Yi Chu writes:

 > Thanks for explaining. But sorry I still do not get the point.  If I have
 > routers A,B and C, connected through an ATM core.  If I set up VCs to
 > connect A and B, B and C only.  Is it a legitimate configuration?  I mean
it
 > is not necessary to have A adjacent to C.  Packet from A to C can still
be
 > routed through B.  Admitted that those packets would traverse in and out
 > the ATM interface on B.  But is there any architectural reason why the
above
 > configuration is prohibited?

no, the configuration is not prohibited and in fact some routers allow
all three interface still to use an ip address from the same ip subnet.

so you could start with default connectivity between your routers and
then start adding more connections dynamically if there is enough
traffic between a pair. i documented this once in an internet draft that
has now expired.

* juha

  ------------------------------------------------------------------------

   Part 1.2    Type: message/rfc822
           Encoding: 7bit