Re: [v6ops] Update to draft-ietf-v6ops-ula

[Nick Buraglio <[e-mail]>]

Excellent feedback, thank you. I suspect that will come after the 6man 6724
draft -00 is released, but I will make some adjustments for the next
version.

nb


On Mon, May 22, 2023 at 7:55 AM Tim Chown <[e-mail]> wrote:

> Hi Nick,
>
> A handful of comments below.
>
> On 18 May 2023, at 15:27, Nick Buraglio <[e-mail]>
> wrote:
>
> In preparation for submitting a draft to 6man to address the details noted
> in draft-ietf-v6ops-ula, we have published an update to this draft,
> available here: https://datatracker.ietf.org/doc/draft-ietf-v6ops-ula/
>
> This addresses several comments on the list and improves some of the
> language to clarify the behaviors. Input is definitely welcome and
> encouraged.
>
>
> One nice quote from 7724 that should be included prominently here is the
> “It is important that implementations provide a way to change the default
> policies as more experience is gained”, as that experience (10+ years) is
> what has led to your draft.
>
> In the abstract you could leave off “The behavior of” at the start, as
> there are many behaviours and the one we’re interested in is preference
> over IPv4 for address selection.
>
> Similarly in the first para of the intro “exhibit opposite behaviour” is a
> little vague, so I’d change that to “have differing priorities in the
> address selection policy table, leading to GUAs being selected over IPv4,
> but IPv4 over ULAs,”
>
> Somewhere, maybe at the end of the introduction, or maybe later on, the
> document should say words to the effect that this differing preference is
> not an issue either if ULAs are not used (which is a pretty common case,
> though we have no sure way of knowing :)) or if ULAs are used in an
> IPv6-only environment (i.e., no IPv4).
>
> The document also doesn’t explicitly state whether global IPv6 addresses
> are deployed. I’d hope they are. But there are various combinations of
> ULAs, global v6, RFC 1918s, global v4, and NATs for either(!) protocol.
>
> There is a DHCPv6 option for the policy table, just that it isn’t exactly
> widely implemented. RFC7078.
>
> Ideally all implementations would support a way to change the default
> policies, the question is whether that can even always be possible.  That’s
> a SHOULD vs MUST that can be discussed through the 6man document.
>
> Tim
>
>
>
>
>