There may be some terminology/semantics that need to be clarified regarding my usage of the term "link". What I meant to convey was that for 3GPP (UMTS), a primary PDP context (and its associated secondary PDP contexts) are treated as a single IPv6 "link". But the GGSN views each primary PDP context as a single "subnet". Therefore, DAD is not required. Similar for 3GPP2 (cdma2000), except PPP is used, etc.
Pete
-----Original Message-----
From: Ole Troan [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, March 25, 2003 1:29 PM
To: Barany, Peter [RICH1:2H16:EXCH]
Cc: Mika Liljeberg; [EMAIL PROTECTED]; [EMAIL PROTECTED];
[EMAIL PROTECTED]
Subject: Re: DAD in node requirements draft
[...]
> During 3GPP (UMTS) work on PDP context activation, it was agreed to make the
> /64 prefix unique per primary (and associated secondary) PDP contexts so
> that DAD would not have to be performed (particularly since privacy
> addresses are specified for use in 3GPP). See RFC 3314 and, I believe, 3G TS
> 23.060 (and/or maybe another document in the 3GPP 24 series).
>
> 3GPP2 (cdma2000) uses PPP and not a PDP context. In IS-835-B it is stated
> that the /64 prefix(es) must be unique to that PPP link ... again avoiding
> DAD when privacy addresses are used.
requiring that a /64 is unique on a link does not avoid DAD.
doesn't the 3GPP specs restrict one end of the link to not use any
addresses from the prefix, if so DAD is not needed.
> Since the IETF IPv6 WG is in the process of updating the IPv6 PPP RFC 2472
> (more specifically IPV6CP), my question is: What changes are being planned
> to that RFC? Do any of these changes pertain to the granularity/uniqueness
> of the /64 prefix? Thanks.
the only thing I can see requiring an update is that the value of
DupAddrDetectTransmits should default to 1 not 0.
you can turn off DAD in your environment if you know you don't need
it.
/ot
