Tim Chown wrote: > If that's static /48's, the /29 boundary will need revision...(and > certainly a /35 would be useless to any medium ISP).
Did the term 'slow-start' get lost somewhere? As I understand rir policy, the /35 was never intended to be the only allocation a serious ISP would get. Static vs dynamic is a non-issue, because either there are enough prefixes to route the currently connected customers or there aren't. If a service provider wants to allocate prefixes to customers that aren't connected, they will need a larger block than the one that allocates dynamically based on connected status. > > Would you apply the RFC3194 0.8 HD ratio to subnets within a > single ISP? > I don't see that provider networks would be flat or near 100% > utilisation > as you suggest in your previous email. Within the context of a pop, why not? The only reason this would not be the case is when a provider considers the allocation to be disjoint from current connected status (ie: static assignment) and the measurement is done on the connected set rather than the allocated set. The HD ratio should be calculated on the allocated set no matter if that is static or dynamic. I can already hear the 'what about renumbering nodes?' question. There is no reason for this to be an issue. Address use is defined to match the smallest scope appropriate for the connection, so intra-site/on-link communications will be immune to changes in the global prefix. If a site is currently disconnected from the service provider, there is no reason for it to have a global prefix floating around. The only reasons for an ISP to make static prefix assignments would be if the site wanted stable DNS entries, but with a decent DDNS infrastructure that is not even required. 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] --------------------------------------------------------------------
