--On Monday, July 1, 2024 16:34 -0400 John Levine <johnl@xxxxxxxxx>
wrote:
> It appears that Ofer Inbar <cos@xxxxxxxxx> said:
>> Are you really saying that the IETF is a special kind of buyer that
>> needs to remain "neutral" in some way, and NOT exert any market
>> force, even if the IETF does have preferences?
>
> The IETF is not a large organization, and it is not a large buyer of
> online or hosting services. If we tell vendors "you must provide
> IPv6 or you won't get our business," if they already have IPv6 for
> their own reasons, OK. Otherwise the answer is "then we won't be
> doing business with you."
>
> We can certainly ask, but we can't force anyone to do anything.
But, coming back to a variation on both Ofer and Christian's points,
it seems to me that we are making a statement by not asking, and that
statement is really not about market forces. It is about whether, if
some organization is considering whether to move rapidly toward IPv6
and asks the question "what is the IETF doing with its own
infrastructure", we want the answer to interpreted as "the message
from the IETF is that IPv6 is not really ready yet". IMO, even
without evidence that a single would-be buyer of services applies
that test, the optics of the IETF not asking for IPv6 support are
terrible and likely to have bad effects on our credibility as a
definer and setter or standards.
It seems to me that the right way to thread that needle is similar or
identical to what David Farmer suggested [1]:
RFPs should require IPv4 support and ask for IPv6 support. A bidder
might reasonably respond with IPv4-only, IPv4 and IPv6, or IPv4 with
IPv6 and some specific marginal cost. As long as we can avoid
trapping ourselves into a strict "low bid wins", situation, we don't
need to tell them how we will weight those considerations and others,
but we at least deliver the message that IPv6 is important to us.
And we do so without shooting ourselves in the foot if it is not
available at all from any vendor who is considered competent on other
dimensions or not available except at a unacceptable marginal price.
I think that, if the service for which an RFP is being issued
involves a particular IETF applications-level protocol, the RFP
should either require conformance or, if that is impractical for some
reason, ask vendors to explain why they cannot or will not conform or
quote a marginal price for conformance. Not really very different.
john
[1]
https://mailarchive.ietf.org/arch/msg/ietf/2XfTTqyUEk_onNA0soAJaH2kuMA
