Re: [dhcwg] Prefix Pool Option for DHCPv6 Relay Agent - Sent @ 3rd

["Bernie Volz (volz)" <[e-mail]>]

This is really very similar to the RAAN
(http://tools.ietf.org/id/draft-ietf-dhc-dhcpv6-agentopt-delegate-04.txt
) work that the DHC WG was working on a while back and stalled. One of
the reasons it stalled was a issue raised by Thomas Narten ... I don't
recall all of the specifics, but I think it had to do with out of order
packets which could end up communicating the wrong information (adding
information after it should have been removed, or vice versa). But I may
be remembering incorrectly - I think I sent an email to the DHC WG
mailing list describing the issue (or Thomas did). This is also an issue
with snooping ... and the SRSN option
(http://tools.ietf.org/id/draft-ietf-dhc-dhcpv6-srsn-option-02.txt) was
there to address some of these issues -- I do vaguely recall Thomas
coming up with another problem that even SRSN didn't solve, but I could
be mis-remembering.

As snooping isn't a DHC WG work item, what exactly various
implementation do to prevent the out-of-ordering issue (which admittedly
isn't very likely but *is* possible) isn't an DHC WG issue.

- Bernie

-----Original Message-----
From: Ted Lemon [mailto:[e-mail]] 
Sent: Monday, October 25, 2010 12:10 PM
To: Ole Troan
Cc: Leaf Y.Yeh; [e-mail]; Bernie Volz (volz)
Subject: Re: [dhcwg] Prefix Pool Option for DHCPv6 Relay Agent - Sent @
3rd

On Oct 25, 2010, at 12:00 PM, Ole Troan wrote:
> no, it wouldn't. the prefix pool option would only get you the
aggregate. the router would still need the more specifics per IA_PD
routes.

Hm, are you sure about that?   I will admit that I haven't read the
draft closely either yet, but I thought the relay option contained the
information needed.   If the relay option is aggregating routes for more
than one client, that would be a problem.

> correct! ;-)
> (this is done when the box is provisioned though, and certainly not
something you'd do every day).

I think that depends on the environment.   In some cases obviously
stability is important, but in the case of something like a mobile
phone, I get the impression that these delegations would be fairly
temporary.

> yes, agree. even though as stated above, you would still need to snoop
for the more specifics.

When we were talking about RAAN, one of the motivations was to avoid
snooping.   It would be relatively easy to have the relay option contain
the actual delegation as well as the enclosing prefix, if that were
desired.