Re: [tcpm] Fwd: New Version Notification for draft-touch-tcpm-experimental-options-00.txt

[Wesley Eddy <[e-mail]>]

On 10/24/2011 5:57 PM, Joe Touch wrote:
> Hi, all,
>
> The following document describes an informal proposal I made on the list
> a while back. Hopefully we can discuss it in Taiwan.
>


After reading and thinking about it, I think we should adopt this.

I think this should be more than just Informational, if people agree
that it's the best way to disambiguate between concurrent experiments,
then I would think that it should be a Best Current Practice.

Some specific comments:

- I believe there are other devices (e.g. Bluecoat) that are also
   using the experimental codepoints (not just the AO precursor and
   the TCP-CT code).

- We might add some analysis that says:

   Hosts (or routers) that have deployed the pre-TCP-AO option using
   the experimental codepoints will generally be limited to BGP-speakers
   or other infrastructure, and not consumer devices or servers at the
   edges of the network that would be participating in other TCP
   experiments.  There is not expected to be an issue with use of this
   nonce for new experimental options to conflict with these.

   Middleboxes deployed using the experimental codepoints for discovering
   each other are expected to pass-through experimental options using the
   nonce discussed in this document, based on discussion with the vendor.
   These middleboxes have their own mechanism for disambiguating their
   use of the codepoints from other uses.

- I think we would encourage new experimental options to strongly make
   use of this nonce mechanism, but say something about the fact that
   if they don't, their odds of causing trouble for other experimental
   options that do use a nonce are small, however the odds of code that
   uses non-nonced   options getting confused while attempting to
   parse nonced options may be greater!  Thus, it's strongly encouraged
   to use a nonced format, and in one's best interest.

- Consider borrowing (or adapting) some of the text from section 1 of
   draft-eddy-tcpm-addl-exp-options-00, in order to motivate why we need
   to encourage this rather than continue to allow the option space to
   be reduced by squatting.

Thanks for writing this!

-- 
Wes Eddy
MTI Systems