Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!zaphod.mps.ohio-state.edu!tut.cis.ohio-state.edu!cica!iuvax!rutgers!att!cbnewsh!tds
From: tds@cbnewsh.ATT.COM (antonio.desimone)
Newsgroups: comp.protocols.tcp-ip
Subject: Re: Traffic Sensitive SPF Routing is NOT too hard!
Message-ID: <6251@cbnewsh.ATT.COM>
Date: 4 Dec 89 18:12:39 GMT
References: <4118@sbcs.sunysb.edu>
Distribution: usa
Organization: AT&T Bell Laboratories
Lines: 36

From article <4118@sbcs.sunysb.edu>, by brnstnd@stealth.acf.nyu.edu (Dan Bernstein):
> In article <6203@cbnewsh.ATT.COM> tds@cbnewsh.ATT.COM (antonio.desimone) writes:
>> An apparent difference
>> is that routing in datagram networks can (in principle) react to
>> congestion on the timescales on which queues build up.
> 
> Since when? I don't know of any methods of controlling flapping and
> instability in principle. When the Internet is highly loaded it shows
> that dynamic routing fails in practice as well.

I'm with you Dan!  You're making exactly the same point I tried to
make later in my posting (did you get there? :-).

> There's absolutely no reason not to use a Bayesian or maximum-entropy
                     ^^^^^^^^^ including being able to calculate routes
                               in a human lifetime?
> calculation of the minimal-cost distribution of paths for a given
> distribution of load. Maximum-entropy routing yields the efficiency of
> dynamic routing without the instability of dynamic routing. If you're
> paranoid, recalculate paths every week instead of every month.
> 
> ---Dan
This is a little too heavy on the jargon for me.  Are you suggesting
a single optimal route?  Or some kind of table-driven routing with
tables calculated in advance?  That's a fine idea (in principle:-) and
has been used in the phone system (*that* again) quite successfully.
It is by no means obvious or trivial to build routing tables, however!
And if "Maximum-entropy routing" means some kind of global optimization
over the space of possible tables, the method does not scale to a large
network.  If it means something more specific, maybe you could send me
(or post) a reference? 
-- 
Tony DeSimone
AT&T Bell Laboratories
Holmdel, NJ 07733
att!tds386e!tds


Brought to you by Super Global Mega Corp .com