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

["Pascal Thubert (pthubert)" <[e-mail]>]

Exactly;

It can be argued that the LLN is a single something that is a more complex form of the classical host-router link from the L3 perspective.
RFC6437 understands that the host can leave the flow label to all zeroes and allows the first router to set it in a recommended fashion. 
The draft pushes that role to the RPL root that is effectively the gateway from the LLN onto the Internet.
Upon the WLLC discussion I have added " Updates: 6437 (if approved) " and posted an update with no other change in the text.

Cheers,

Pascal


> -----Original Message-----
> From: ipv6 [mailto:[e-mail]] On Behalf Of Michael Richardson
> Sent: lundi 4 août 2014 16:15
> To: Ole Troan
> Cc: roll; 6man WG
> Subject: Re: WGLC for draft-thubert-6man-flow-label-for-rpl-03
> 
> 
> On 01 Aug 2014, at 21:58 , Brian E Carpenter <[e-mail]>
> wrote:
>     >> It is true that the RFC also says
>     >>
>     >> A forwarding node MUST either leave a non-zero flow label value
>     >> unchanged or change it only for compelling operational security
>     >> reasons...
>     >>
>     >> There is a case for arguing that flow-label-for-rpl should carry the
>     >> legend "Updates: 6437 (if approved)" because it adds another
> exception
>     >> to this sentence. There is also a case for arguing that
>     >> flow-label-for-rpl is a stateful mechanism, out of scope for RFC 6437
>     >> according to Section 4 of that RFC.
>     >>
>     >> Either way, I don't see a problem.
> 
> Ole Troan <[e-mail]> wrote:
>     > I see the principles in these two documents as mutually exclusive:
>     > RFC6437: middleboxes can mess with the flow label -flow-label-for-rpl:
>     > networks uses flow label field as an immutable protocol field
> 
>     > you can't both use the flow label as an immutable protocol field and
>     > have middleboxes rewriting it.
> 
> This is my take.
> Case 1:  packet is within the LLN.  No issue; the sender sets the flow label,
>      and we know (by design) that there aren't any (non-RPL) middle boxes
>      that need to rewrite it.
> 
> Case 2: the packet has to leave the LLN, travel across the Internet, and
>      enter another LLN.
>      Here the arguments about the LLN being a kind of virtual host from
>      the document are relevant.  The flow label, when it crosses the
>      Internet, would be immutable.  Prior to this document, the flow label
>      would have been compressed out, been zero, and the DODAG root ought
> to
>      have set it in some consistent fashion anyway.
> 
> --
> Michael Richardson <[e-mail]>, Sandelman Software Works
> -= IPv6 IoT consulting =-
> 
>