[1526] in Commercialization & Privatization of the Internet
Alternate Routing
daemon@ATHENA.MIT.EDU (Stephen Wolff)
Tue Oct 22 08:55:46 1991
To: yakov@watson.ibm.com
Cc: com-priv@psi.com
Date: Tue, 22 Oct 91 08:43:22 EDT
From: Stephen Wolff <steve@ncri.cise.nsf.gov>
> The problem you are trying to solve is to discriminate between commercial
> and non-commercial traffic. So, you need only 1 bit (which is QUITE less
> than the number of bits needed to specify a network).
AHA!! That may be YOUR problem, Yakov, but it's only PART of mine. I need
also to discriminate, for example, packets between two NASA Centers from
those only sourced at a NASA Center from those only sunk at a NASA Center.
And (hypothetically) Albanian packets from Soviet Packets from Hungarian
packets. Also packets sourced or sunk at Project MegaBux (entitled to use
the REALLY fat pipes) from the routine FLASH OVERRIDE stuff. Etc., etc.,
etc. ...
Well, sure, my tongue was pretty firmly planted in my cheek in my earlier
posting, but we really do need either source-specific routing, or some modest
number of bits (but > 1) for ToS. And since the requirements for "special"
routing - of whatever sort - are not immortal, the meaning of those few ToS
bits must be allowed to change and the new meanings understood throughout the
portions of the Internet which MAY be affected - on some relatively long time
scale. Like seconds. When requirements are viewed this way, source-specific
routing may in fact offer greater efficiency than overloading the ToS bit(s)...
-s
