Hi Tony and Noel,

Tony, thanks for your responses to my message.

If you want to stick with a DFZ definition which includes or is
restricted to ASes, then Noel's explanation, my previous
explanation, my new one below are inappropriate.

However, in the RRG we are primarily or solely concerned with
routers - specifically "DFZ routers" which are the devices
affected by the scaling problem we need to solve.  So I think it
would be better to have a definition of "DFZ" as it relates to
RRG work.

If you think a router-specific definition is in order, here is a
shorter suggestion prompted by Noel's suggestion:

> I agree we could use a better DFZ definition, but I find your proposed
> definition to be too obscure and technical. I would prefer something in
> simpler English, like "the set of Internet routers which contain full routing
> tables, i.e. contain entries for all available destinations in the Internet".
> One could also add something about how "the DFZ does not contain those
> routers with a default route, potentially augmented with a limited subset of
> the full routing table", or something.

  DFZ - Default-Free Zone:

   The set of routers which need routes for all, or almost all, the
   prefixes advertised in the interdomain routing system.  Routers
   with a single "upstream" link to all ASes other than their own
   can use a default route to cover most of these prefixes.

This definition explains the function of the default route, and by
implication the reason for the rather curious term "Default-Free Zone".

I think this definition is more precise, by using "need", since the
definition suggested by Noel would also include routers with a single
upstream link, which could use a default route, but were instead
configured to have the full routing table.

If a Route Reflector with a single "upstream" link is to be considered
as a DFZ router, please add "Many" at the start of the second sentence
to allow for these without adding another two sentences of fine-print.

Perhaps you can delete ", or almost all,".


  - Robin

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

Reply via email to