Re: Eric Rescorla's Discuss on draft-ietf-httpbis-cdn-loop-01: (with DISCUSS and COMMENT)

Well, this argument has some force, but my point is that this isn't
substantively different from the initial version of the draft in this
respect.

-Ekr


On Sun, Jan 27, 2019 at 2:10 PM Mark Nottingham <mnot@mnot.net> wrote:

> HTTP has used similar constructs for 20+ years and collisions have not
> been an issue:
>
> https://httpwg.org/specs/rfc7230.html#header.via
> https://httpwg.org/specs/rfc7231.html#header.user-agent
> https://httpwg.org/specs/rfc7231.html#header.server
>
> All of them address the issue of pseudonym collision by recognising that
> implementers have an interest in distinguishing their implementations from
> others if they want to rely upon a field for debugging (as all of these
> are, effectively). None of them addresses the case of an active,
> *intentional* collision (as we see sometimes with User-Agent), because
> doing so would require far too much infrastructure, compared to any benefit
> seen.
>
> Notably, doing something like rooting product tokens in the DNS name space
> would have done *nothing* to prevent the issues we're seeing with
> User-Agent. Doing so would have also brought about the usual questions
> around whether something needs to listen at that hostname, whether it needs
> to resolve, and what happens when a name changes hands.
>
> The failure case of a collision is also pretty straightforward; if I run a
> CDN and another CDN usurps my value, they're only shooting themselves in
> the foot, because it'll look like they're in a loop if I'm configured to
> forward traffic to them, and they'll start refusing requests -- thereby
> screwing their customers. Their natural incentive is to avoid this
> situation.
>
> So, the properties you're looking for are already naturally emergent from
> the documented protocol. While we could require a hostname, I think doing
> so is effectively cargo-cult protocol design; it won't actually improve
> collision avoidance in practice.
>
>
>
> > On 28 Jan 2019, at 6:44 am, Eric Rescorla <ekr@rtfm.com> wrote:
> >
> > Unless I am misreading this text, it seems to not provide any guidance
> at all about how the pseudonyms are constructed, so I don't really see how
> it addresses the issue of pseudonym collision
> >
> > On Thu, Dec 20, 2018 at 9:04 PM Mark Nottingham <mnot@mnot.net> wrote:
> > See:
> >   https://github.com/httpwg/http-extensions/commit/59505dbd1883
> >
> >
> > > On 21 Dec 2018, at 3:34 pm, Adam Roach <adam@nostrum.com> wrote:
> > >
> > > On 12/20/18 7:19 PM, Mark Nottingham wrote:
> > >> On 21 Dec 2018, at 11:40 am, Adam Roach <adam@nostrum.com> wrote:
> > >>> I agree with EKR here. The more I think about the situation -- and
> Mark's responses in particular ("There are relatively few CDNs, and they're
> pretty aware of each other") -- the more I'm convinced that, while the
> proposed solution *probably* works in 2018, it may well fail in 2028,
> depending on how the CDN market evolves. If history is any guide, engineers
> are pretty terrible futurists. The best we can do is plan for scaling
> beyond that which we can presently imagine (cf. IPv4).
> > >>>
> > >>> I think some means of collision avoidance here is required prior to
> document publication.
> > >> I agree engineers are terrible futurists; however (and perhaps as a
> result), we're really really good at over-engineering things for
> anticipated problems that don't eventuate. My viewpoint here is informed by
> observing how many Web-related registries operate (or fail to).
> > >
> > >
> > > No one here is asking for a registry.
> > >
> > >
> > >> The current specification allows us to insert more stringent
> requirements in the future if necessary. Personally, I think that's
> sufficient.
> > >>
> > >> However, if you insist, I think the lowest-impact way to address this
> is to adopt a structure similar to that used in Via, (hostname under
> control of the CDN or pseudonym) and encourage (but not require) use of a
> hostname.
> > >
> > >
> > > It's probably enough to say that it has to end in a domain name under
> control of the CDN, but your proposal works as well.
> > >
> > >
> > >> If we do that, we might want to revisit the example, since its use of
> 'host' might be confusing to readers.
> > >
> > >
> > > I always assume that spec changes that impact examples imply changes
> in the corresponding examples, so yes. Thanks.
> > >
> > >
> > > /a
> > >
> > >
> >
> > --
> > Mark Nottingham   https://www.mnot.net/
> >
>
> --
> Mark Nottingham   https://www.mnot.net/
>
>