Re: [spring] Conclusion from WG poll on dataplane solution for compressing segment routing over IPv6

Hi Tony,

May I ask if the "check" you are recommending is to be done only based on
reading a few drafts or maybe also something else ?

I personally think that while reading the specs may give some hints as to
how the solution is supposed to work the real crux of the matter is in
details of hardware implementations and various design choices made there.

So to me if this WG is to make some choices (which I am not sure if it
should or is capable of specifically in regards to hardware impact of any
solution) much more useful is to document actual shipping and deployed
implementations and support it with actual data from the field (both
production deployment as well as early field trials).

Data presented so far clearly indicates that CSID data plane with two
flavors has most vendor and operational experience behind it.

Meanwhile let's see if the below two drafts will be adopted as WG doc:

https://datatracker.ietf.org/doc/html/draft-srcompdt-spring-compression-requirement

https://datatracker.ietf.org/doc/html/draft-srcompdt-spring-compression-analysis

If they will IMHO this will be good technical as well as formal input to
the choice to be made.

Many thx,

Robert

On Wed, Sep 8, 2021 at 12:58 AM Tony Li <tony.li@tony.li> wrote:

>
>
> Dear chairs & WG,
>
>
> > Our thanks to the working group members for speaking up clearly.  There
> is a rough (quite clear) consensus for standardizing one dataplane solution
> to compressing segment routing over IPv6.
>
>
> I’m very pleased that we’ve found the courage to say that we want to make
> a decision.
>
>
> > As chairs, there are some related observations we need to make.
> > There appears to be significant interest in using the framework in the
> CSID draft for addressing the above.
>
>
> I concur that there is interest in CSID.  However, as we have not yet done
> a consensus check to select the specific proposal, I hope that this is not
> your assertion that CSID is the result. This is a major decision by the WG
> and it does deserve a formal and explicit check.
>
>
> > However, before we issue a call for adoption on that, the chairs would
> like to understand how the working group wants to solve a technical
> problem.  The CSID draft contains two dataplane solutions.  The above rough
> consensus is for one dataplane solution.  Does the working group want to
> choose one?  Do the authors want to suggest that one of the two is the one
> we should standardize, and get working group agreement?
> > Should we adopt the document, with a note indicating the problem, and
> solve the problem afterwards?  (That itself does not solve the problem, it
> merely kicks it down the road.) Do folks see another means to avoid putting
> the WG in conflict with itself?
>
>
> If the authors of the CSID proposal, or any proposal, would like to revise
> their proposals before we make a selection, this would seem to be an
> opportune time.
>
> Tony
>
> _______________________________________________
> spring mailing list
> spring@ietf.org
> https://www.ietf.org/mailman/listinfo/spring
>