Behaviour and consequences of DAD

Mathew Newton mathew at newtonnet.co.uk
Wed Oct 16 21:11:53 CEST 2013


Hopefully the group won't misconstrue the subject line as being a plea for
domestic assistance!

I am trying to understand the nuances of how DAD operates, and in
particular the consequences of how it appears to work (to me at least).

Specifically, it is my undertanding that when an node has a tentative
address it wishes to use it sends a neighbour solicitation to the
solicited-node multicast address which is computed using the last 24 bits
of that tentative address. If another node is already listening on that
solicited-node multicast address it will reply and the duplicate detected
(and avoided).

To me, that therefore means that the effective number of discrete
addresses on a link can only ever be a maximum of 2^24 (65,536). Is that
correct?

To put it another way; two *different* addresses can result in the same
solicited-node multicast address given that only the last 24 bits affect
the outcome and therefore the potential size of unique address space is
reduced.

I am of course only wondering this from a theoretical perspective given
that 2^24 is still a large number however there does seem a general
assumption that a /64 subnet could contain upto 2^64 devices.

I hope my rationale makes sense, and that you can either point out the
error of my understanding or confirm if it is indeed valid.

Regards,

Mathew




More information about the ipv6-ops mailing list