I haven't been able to join the jabber session, but have been
listening to the audio.

Lixia said that Ivip requires changes to routers other than "edge
routers".  I think she meant DFZ routers.  This is only for "Modified
Header Forwarding" as an alternative to encapsulation.  This is
purely optional and is for the the long-term future.  So it is not
not at all required for Ivip to work well.

She also said that Ivip floods all mapping changes, implicitly
everywhere.  The new Distributed Real Time Mapping system involves
potentially large numbers (dozens to hundreds) of independent systems
of DITRs (Default ITRs in the DFZ), where each system generally has
DITR sites widely distributed around the Net.  Each such system
pushes, in real time, mapping changes to its sites for the MABs
(Mapped Address Blocks) that system handles.

This is arguably "flooding", but it is within the one DITR system and
so only concerns the one or few organisations which run this system.
 There are only likely to be at most a few dozen such sites in each
such system.  Each such system only handles a subset of all MABs.  So
this should be perfectly practical and scalable.

ITRs which are handling traffic packets for a given micronet will
still get the updates they need, but these are directed updates.  The
query servers at the DITR sites send updates to the QSR servers
(typically at ISPs) which need to get them, and these QSRs pass this
on to the ITRs.  This cannot be described as "flooding".  ITRs and
QSRs are purely caching devices.

Please take a look at the text and three diagrams at:

  http://www.firstpr.com.au/ip/ivip/drtm/

DRTM could also be used by other CES systems, such as LISP.


  - Robin

_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg

Reply via email to