On 8/11/2015 10:34 AM, Juliusz Chroboczek wrote: > I'm removing from CC the people who I know are on Homenet, please do the > same with the other lists. > >> Since RFC 3819 is mostly concerned about avoiding receiving unwanted >> multicast, >> >> I don't know why you would get that impression. I helped write that >> section (as noted in sec 19). Native L2 multicast should be supported >> exactly because it's useful to IP, as noted here: >> >> --- >> Multicasting is considerably more efficient when a subnetwork >> explicitly supports it. >> --- > > Perhaps I misread this section,
Are you referring to RFC3819? > but the impression I get is that it is > concerned about avoiding waking nodes and avoiding duplicate > transmissions. This is incorrect. The section is basically saying "IP works better when L2 supports multicast so L3 (IP) doesn't have to emulate it". The assumption is that L2 will do a reasonably good and efficient job of multicast/broadcast - certainly better than L3 or other layers would. > I doesn't read like the authors are thinking about > networks where multicast has higher packet loss and lower throughput than > unicast, which is what concerns us here. The point of that section is that IP is more complex when it has to emulate something that L2 doesn't provide. If it's more efficient to implement L2 multicast via serial copy than by shared channels, that's L2's decision. The point of the RFC is to NOT make this the job of L3. L3 is a bad place to support multicast or broadcast on subnets. Joe _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
