Re: Martha's comments

[Noel Chiappa <[e-mail]>]

    On a negative side by choosing "distribute restrictions"
    rather than "restrict distribution" paradigm IDPR forces
    domains to participate in processing and distribution of
    information that may be of no use to them; thus placing
    an unjustifiable burden on their resources.

	I don't see how IDPR is any worse than any DV protocol here. In
all DV protocols, nodes which are intermediate between a source and
destination have to compute routes which the intermediate nodes
themselves may not be using. How does this differ? In fact, source
routed LS might be better, since intermediate nodes don't calculate
routes at all unless they themselves need them.


    Thus, while the architecture can provide information about
    heterogeneous charging schemes supported by various domains,
    the architecture can not realistically support computation
    of a minimum cost route unless all the domains throughout
    the Internet employ the same charging scheme.

	I think this claim (and some of the preceeeding text) is too
broad to be supportable. Most research in graph theory has concentrated
on algorithms which compute optimal routes to all destinations, not on a
pair-wise basis. I think it would be appropriate to wait until the
research is done on the latter case, and see what results. Also, I
suspect that further research on algorithms which are close to optimal
(probably probabilstic, and thus not guaranteed to be within bounds), but
not optimal, will give us algorithms that are much less expensive than
exponential. (Once again, a benefit of source-routed LS; new, improved
routing algorithms can be deployed incrementally.)


    I would suggest you to look at the storage complexity overhead
    for IDRP that was published in the April's issue of CCR.

Is this material available online?


    With respect to the computational overhead, as you pointed out
    further in your message, even with BGP you are likely to get less
    computational overhead than with IDPR.

You have restated something she said in much more sweeping terms than she
did (I refrain with great difficulty from speculating why). She said that
that was true, for route generation only, in certain limited cases. She
then pointed out that:

    If this is not the case, then it is not easy to make a general
    statement about which approach provides lower-cost route generation.

She also pointed out that for all phases except route generation:

    With these not unreasonable assumptions, I was somewhat
    surprised to find little difference between the two protocols
    in the storage, communications, and computation required for
    distributing, processing, and storing routing information and
    setting up and storing forwarding information.  ...  Instead,
    I found that the costs were very close and in several cases
    IDPR was even less expensive than BGP.



	Noel