Re: [spring] Spirit and Letter of the Law (was: Question about SRv6 Insert function)

[Fernando Gont <[e-mail]>]

On 7/9/19 04:07, Mark Smith wrote:
> 
> 
> On Sat, 7 Sep 2019, 09:00 Tom Herbert, <[e-mail]
> <mailto:[e-mail]>> wrote:
> 
>     On Fri, Sep 6, 2019 at 3:11 PM Robert Raszuk <[e-mail]
>     <mailto:[e-mail]>> wrote:
>     >
>     > Sander,
>     >
>     > But this is exactly what both chairs of 6man did with the help of
>     AD long time back. You must have missed it !
>     >
>     > And below is a link precisely written to address requirement of
>     justifying deviation:
>     >
>     >
>     https://tools.ietf.org/html/draft-voyer-6man-extension-header-insertion-06
>     >
> 
>     Yes, and that draft was discussed in depth in 6man, a number of issues
>     were raised, and we haven't heard back from the authors. And this was
>     not just a case of saying it violates RFC8200, there were specific
>     reasons given why it isn't robust. So it's correct to say that current
>     consensus in 6man is that extension header insertion is disallowed.

Exactly. The relevant text in RFC8200 was added *exactly because of
this*, indeed.


> 
>     I also have a nit about that draft. The very first line of the
>     abstract is "The network operator and vendor community has clearly
>     indicated that IPv6 header insertion is useful and required." The fact
>     that somebody _really_ wants their protocol to be adopted by IETF is
>     hardly going to influence anyone. Obviously, anyone writing a draft
>     thinks that it is useful and required, it's up to the authors to show
>     that. Frankly, I think this statement comes off as being condescending
>     and sets a poor tone for reading the rest of the document.
> 
> Yes, it's a appeal to authority.
> 
> Earlier versions had 10 or more authors listed. That is also an appeal
> to authority.

https://tools.ietf.org/html/draft-filsfils-spring-net-pgm-extension-srv6-usid-00
has 26. About 5 authors per non-boilerplate page.



> The draft doesn't say why insertion is considered necessary. There is no
> justification presented. 
> 
> There is no statement that says, "When using IPv6 tunnelling with 128
> bit SIDs, the per packet overhead can become too high." That would be
> admitting what the real cause was, and make people question whether
> using 128 bit SIDs was really the right decision - as they should.
> 
> Considering all that, I would describe this draft as trying to get its
> proposal approved through social engineering rather than technical
> engineering.

Indeed. The original argument was "we do it, because it is not
forbidden" -- when I think it was obvious to everyone that IPv6 didn't
allow for header insertion.

Then there was a big push to keep what became RFC8200 ambiguous in this
respect.

Eventually we got IETF consensus to incorporate specific text that
forbids EH insertion, for all the considerations that had been made
while the topic was discussed in 6man.

Now the argument is that we should kind of ignore RFC8200, and "judge
the technical merits of EH-insertion", as if the text we included in
RFC8200 was just some random bits we added into the spec, because we had
nothing else to do.

And having us to find if EH-insertion breaks anything (which was already
discussed ad nauseam on 6man), rather than the proponents making a case
about why they need EH insertion in the first place.


-- 
Fernando Gont
SI6 Networks
e-mail: [e-mail]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492