> So we have a site with a globally-routable /48 prefix. Are you saying > that instead of using site-local addresses in addition to its global > addresses, the site could use a second /48 global prefix that is not > globally routable and is filtered at the site boundary?
absolutely. > This would be silly. It would be just like site-locals except with a > non-standard prefix. no, the prefix should be standardized. > It would work worse than site-locals because > Default Address Selection wouldn't work properly Default address selection doesn't work properly anyway, partially because it favors SLs over globals, and partially because the idea that you can expect a default address selection rule to work in the absence of information about the app's requirements, and to do the right things with scoped addresses, is naive. > and because of Keith's > issue of distributed apps that do referrals - those apps wouldn't be > able to distinguish the two kinds of addresses to do the right thing. sure they would be able to distinguish them, because the prefix would be standardized. however "the right thing" is simply to favor routable globals over non-routable globals but to try all of them - that way, if two nodes don't manage to connect, it's either because the net is broken or the net is prohibiting the connection - either way it's beyond the the app's control. the app is able to do everything that can possibly be done and with almost zero additional complexity. compare this to the SL case where the app either has to a) artifically limit connectivity by filtering SLs when it's not sure that nodes could connect using them, OR b) include potentially SLs in referrals define an addressing scheme for all nodes used by the app whenever connecting to an SL address check to see whether you're really connecting to the node you think you're connecting to and perhaps implement routing across SL boundaries for the case where the customer demands that the app support it (which is quite often the case - people do expect apps to work across site boundaries) -------------------------------------------------------------------- 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] --------------------------------------------------------------------