In your letter dated Wed, 25 May 2011 08:05:27 -0700 you wrote: >I said that the current assumed maximum 3 NUD probes one second apart >makes IPv6/ND less flexible than IPv4/ARP, and I think we should remove >those constraints. > >I'm having a hard time understanding what point you are trying to make, >and whether or not it relates to the text in the draft. Please send some >suggested text changes to the draft if you think that would address your >concern.
In my opinion, the current spec is simple and it works well. The advantage of precisely defining all the protocol contants also means that you predict what nodes are going to do. With ARP you are never sure. One way to move forward is to define an option for router advertisements to change some of the constants you want changed. Another option is to just update the RFC with new constants. NUD a lot of stuff. It detects that a system is down, allowing higher levels to take an alternative path. It detects that a system got a new MAC address (something you can easily found out from ARP broadcasts, but not as easily from ND multicasts), it can expire routers that down, not just for default routers but also after a redirect, and for prefixes with the onlink bit clear, the same applies to hosts. So there is a lot of stuff going on there. You do really want to just randomly mess with that? >From reading your draft, you never seemed to have considered the case that a node would get a new ethernet address. So, I have two complaints. The problem statement is one of hand waving. Yes, there will be a few more multicasts. No numbers on how badly it fails. Second, I don't think it should be left to individual implementors. If you want to create an additional nob for administrators, then just do so. Or if 3 NUD packets is just too few, then propose increasing it. -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
