Le 1 août 2011 à 17:04, Qiong a écrit : ... > So, this is a problem about how to define appropriate port set for our > customers, or to define maximum concurrent subscribers for a given IPv4 > address pool. Otherwise, there would either be a waste of resource, or port > exhaustion.
> Maybe we can even make some more flexible port-set rules for different > time-slot or different types of users. There is AFAIK no need to make port-set rules themselves more flexible. Two customer classes (and tariffs) with standard port-set sizes should IMHO be sufficient: - One for exclusive IOPv4 addresses - One for shared addresses. Their different port-set sizes are expressed by the difference between lengths of their IPv6 assigned prefixes. Three classes would give even more flexibility, but it doesn't seem to me necessary. (Keep it simple, stupid!) > But anyway, this is the problem we have to face in the further, and > I also suggest that port-consuming applications should upgrade to IPv6 > directly. Absolutely. Actually, where native IPv6 addresses are already available, a significant part of the traffic is already IPv6. Regards, RD _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
