Re: Please review: 6MAN WG Adoption call: draft-baker-6man-hbh-header-handling-03

["Bless, Roland (TM)" <[e-mail]>]

Hi,

Am 18.10.2015 um 20:47 schrieb [e-mail]:
> This message starts a two week 6MAN Working Group Last Call on adopting:
> 
>   Title           : IPv6 Hop-by-Hop Header Handling
>   Authors     : F. Baker
>   Filename   : draft-baker-6man-hbh-header-handling-03
>   Pages       :
>   Date          : 2015-10-6
> 
>   https://tools.ietf.org/html/draft-baker-6man-hbh-header-handling-03
> 
> as a working group item. Substantive comments and statements of support for adopting this document should be directed to the mailing list.  Editorial suggestions can be sent to the authors.  This adoption call will end on November 1st, 2015.

I support adoption of the draft.

Some comments:
   Router Alert Option:  The IPv6 Router Alert Option [RFC2711]
      [RFC6398] is intended to force the punting of a datagram to
      software, in cases in which RSVP or other protocols need that to
      happen.

"to software" could be more precise, e.g., control plane software.
Usually, RAO is used to intercept signaling (control) packets. Moreover,
since a 16-bit value in the RAO header is used as codepoint for
the signaling application, the action may depend on the particular
value.

>> 2.1.  Hop-by_hop Options

>>   At this writing, there are several defined Hop-by-Hop options:

Since RFC 2460 only allows a single hbh extension header
within an IPv6 packet, the listed options cannot be used
at the same time, right?
However, I can consider scenarios where it makes sense to
actually combine those.

>>  2.2.  Changing options in transit

This is valuable in order to provide more explicit feedback for
congestion control as already mentioned.

>> 2.3.  Adding headers or options in transit

Indeed this may have a lot of difficult implications, so we need
to discuss whether this makes sense. Moreover, there is nothing said
about practical difficulties with the keeping the processing speed
at line rate when stuff is inserted. While I know that encapsulation
can be done at high speeds, I'm not sure about insertion of headers....

>> 5.  Security Considerations

I think that this needs more work and that at least Section 3
of RFC 6398 https://tools.ietf.org/html/rfc6398#section-3 is highly
relevant here. RAO packets are usually punted to the slow path
since it is the control plane. While it is clear that
draft-baker-6man-hbh-header-handling-03 wants to at least skip
unknown or unconfigured hbh extension headers at line speed,
I'm not sure that the control plane processing speed can
be that high...

Regards,
 Roland