Hi, everyone. We have just uploaded a new draft of TCP-ENO. Sorry for
the delay. The new draft incorporates the feedback from the last IETF
meeting and the mailing list. Here is a summary of the changes:
* A SYN segment can now contain at most one TCP-ENO option, because
order matters and people were worried about middleboxes rearranging
TCP options, and adding logic to sort different options seemed like
it would add complexity.
* There's now a concept of a marker byte that specifies non-zero data
length for the following TCP-ENO suboption. Hence, more than one
suboption can contain data. However, the optimization remains that
the last suboption to contain data does not require a marker byte.
* The definition of role priority has been changed so that the 'b' bit
is now more significant than the 'p' bit. This is in response to a
point made by Bryan Ford that because of dropped SYN segments,
potentially only one side of a TCP connection knows it is
participating in a simultaneous open. We also added language
discussing this case.
* As requested at the last meeting, the new draft clarifies that it is
the responsibility of encryption specs to authenticate any data in
non-SYN ENO options. (Such data is optional and ignored by ENO.)
* The subsection of section 6 pertaining to simultaneous open has been
removed, since despite a small number of objections, the working
group seems to have settled on TCP-ENO's design point (i.e., TCP
never fails, but encryption requires application help for
simultaneous open).
* The subsection of section 6 dealing with optimizing suboption data
has been removed, now that there is a marker byte mechanism that
essentially offers the best of both worlds.
The new version is available here:
https://datatracker.ietf.org/doc/draft-ietf-tcpinc-tcpeno/
We believe this does not change very much for the encryption spec
drafts, other than the citation. We plan to submit an updated version
of tcpcrypt very soon, but with possibly only one word change and a
citation update.
We will also update the informational API draft within the next few
days.
David
_______________________________________________
Tcpinc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tcpinc
