Hi, I was asked to do an early review of the i2rs-rib-info-model...
I liked the comprehensive approach describing the RIB, including tunnels, multi-topology routing (by using multiple RIBs) and routing reactions (like drop/icmp-error). I found a few things in the draft that in my opinion need a bit more work... First it seems that Section 2.3 (Route) is a bit out of sync with the BNF later in the document, it should at least mention matching the source-IP address of the IP headers. Second (if I read the BNF in Section 6 correctly), the match for a route seems to be one of the list "ip address, MPLS label, MAC address, interface". I think it should be possible to combine "interface" or "mac address" with an IP address to restrict the focus of a route, e.g. "match fe80::1 from interface X". Last, I wonder if multicast routing needs more different types of matchers, e.g. a match on the TTL of the IP packet to limit the range of a multicast group. There is also problem of multicast routing in MANETs (see RFC 6621) which can use a hash-based duplicate detection to determine if it forwards or drops a multicast packet. Would this be out of scope for the draft? Henning Rogge _______________________________________________ i2rs mailing list [email protected] https://www.ietf.org/mailman/listinfo/i2rs
