Re: [Roll] [6lo] WGLC for draft-thubert-6man-flow-label-for-rpl-03

[Carsten Bormann <[e-mail]>]

>> This seems to me to be the strong technical argument in favor of the draft. While Carsten's approach seems better technically, it only works for 6lowpan, and RPL should be independent of the underlying link layer.

The transport of RPL specific information for data plane packets already *is* independent of the underlying link layer, as specified in RFC 6553 and RFC 6554.  All we are talking about here is an optimization for those (for 6553, specifically; actually 6554 is probably the one even more in need).

> Could we consider Carsten's approach a compression mechanism for the RPL information options, rather than a replacement, thereby allowing a one-to-one translation between compressed mode on 6lowpan networks and uncompressed mode in other networks?

Indeed, that is probably the best way to view this approach.
(In the next version of the draft, I’ll emphasize this.)

My assumption here is:
— Any subnet that could not live with transporting RFC 6553 as is, i.e. 8 bytes extra compared to the flow label approach (for uncompressed packets*), already needs (and, at this point, has) a header compression solution for the expansive IPv6 header anyway.
  So adding RFC 6553 compression to that solution is the natural approach.
— All subnets that can live with the IPv6 header as is, also can live with RFC 6553 as is.

This argument is on a technical level.
I’m not talking about cognitive dissonance**) here, or political issues.
If there are important political issues (I have no idea what the ISA100/WCI talk is about), they probably need to be brought out in the open here.

Grüße, Carsten

*) less difference for header-compressed packets, e.g., 5 bytes for 6lo.
**) http://www.iab.org/wp-content/IAB-uploads/2011/04/Bormann.pdf, section 2.6