Keith Moore wrote: > ... > I'd be happy to see a scalable alternative to provider-based > addressing, > but that's not a good argument against SLs with site-ids.
I was not arguing against SLs with site-ids, just that we should not try in any way to lead people down the path where those site-ids are perceived to be globally unique. As long as the site-id is a locally administered value, a network administrator can use them privately in any way he sees fit, including private connections to other networks (assuming they can coordinate amongst themselves to avoid collisions). My primary concern is that we avoid building what amounts to an address registry of global scope, which ends up maximizes entropy in the routing system. Yes we need PI public space, see: draft-hain-ipv6-pi-addr-use-02.txt for some of the reasons why. The actual mechanism is still to be resolved, but the fundemental need remains. We also need space that is set aside for private use, and the SL space does that with the exception that it tried to legislate that /16-/48 must be 0. If I were building a network, since FEC0: is defined for local use, I see no reason that I would leave those bits as 0. Tony -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------
