Re: <draft-ietf-6man-default-iids> update to rfc2464bis

[Tim Chown <[e-mail]>]

Hi,

On 12 Jan 2017, at 08:24, Lorenzo Colitti <[e-mail]<mailto:[e-mail]>> wrote:

On Thu, Jan 12, 2017 at 4:20 PM, Fernando Gont <[e-mail]<mailto:[e-mail]>> wrote:
> We do
> recommended that nodes use RFC7217 addresses by default, but only if the
> node wishes to create stable addresses, and there is no requirement that
> a node do so.

So far, the current specs essentially require that. RFC4941 (the only
spec we have for non-temporary addresses) require that the be generated
along with the traditional (stable) addresses.

I don't see what RFC 4941 has to do with this discussion. RFC4941 is a method of generating temporary addresses. It is not required to be used, and it is not the only way of forming IP addresses that can change over time.

Indeed, it’s orthogonal to this discussion.

FWIW, I do think there are scenarios in which you might want to do
temporary-only, but we certainly need an update to the current specs in
that regard (and guidance regarding where to use each).

There are lots of ways of forming interface identifiers. We should absolutely not rule out the ability to form IIDs from non-stable MAC addresses (IIRC the text in default-iids was carefully modified to ensure that that would still be possible, based on substantial discussions in the WG), and we should *absolutely* not say that the only possible form of interface identifiers on an Ethernet interface are RFC7217 and RFC4941.

Indeed, I was at CERN yesterday and my address (and thus IID) there was assigned via DHCPv6, which is the only way you can get IPv6 addresses there.  RFC2464 pre-dates DHCPv6; should reference to it be included in the -bis?

That aside, in the context of temp-only addresses, doing Modified-EUI64
based on a randomized MAC address is still a bad idea:

I am not going to rehash that argument here. Everything we had to say about that is in the archives.

I support Lorenzo’s suggestion to change the text in the -bis update to emphasise the recommendation is for *stable* identifiers.

This reflects what the default-iids draft clearly states:

"This document changes the recommended default IID generation scheme
   for cases where SLAAC is used to generate a stable IPv6 address.  It
   recommends using the mechanism specified in RFC7217 in such cases,
   and recommends against embedding stable link-layer addresses in IPv6
   Interface Identifiers.”

And it’s only a recommendation. The text does not preclude use of embedded stable MAC addresses if someone really wants to do that.  Nor does it say anything about the relationship to temporary (non-stable) addresses.

Tim