Re: [IPv6] Flow label BCP [was: looking for feedback to draft-eckert-6man-qos-exthdr-discuss]

[Mark Smith <[e-mail]>]

On Tue, 12 Mar 2024, 03:04 Dale W. Carder, <[e-mail]> wrote:

> Thus spake Brian E Carpenter ([e-mail]) on Fri, Mar
> 08, 2024 at 08:40:03AM +1300:
> > Tom,
> >
> > One final point. You wrote:
> >
> > > There is a question as to whether the flow label needs to be fixed for
> the lifetime of a connection.
> >
> > RFC 6437 says:
> >
> > "The 20-bit Flow Label field in the IPv6 header [RFC2460] is used by a
> > node to label packets of a flow."
> >
> > It was intentional to leave the definition of "flow" rather open, because
> > we didn't know what usage might evolve. But the text also says:
> >
> > "From an upper-layer viewpoint, a flow could consist of all
> > packets in one direction of a specific transport connection or media
> > stream.  However, a flow is not necessarily 1:1 mapped to a transport
> > connection."
>
> Sorry I haven't been completely following this thread, but in case folks
> have not seen it, here is a use case of changing the flow label as a
> means of "asking" an ecmp implementation w/in the limited domain to
> re-hash the flow somewhere else when congestion is detected.
>

A fundamental drawback of that idea is that it is making the Flow Label
field name inaccurate, because the same single flow can end up with
multiple different label values over time.

A more accurate name for how the field is being used would be something
like "Traffic Engineering Label".



> https://www.cs.utexas.edu/~venkatar/sys_perf_analysis/plb.pdf
>
> Dale
>
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> [e-mail]
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------
>