Re: [Idr] WG adoption for draft-haas-flowspec-capability-bits - 3/30 to 4/13

[Robert Raszuk <[e-mail]>]

Jeff,

Before answering first I wanted to get clarity on question #1. Now I have
it. Hence ready to answer the questions:

> 1) Is this document clear about the proposal?

No. Current text gives too much freedom to implementation in what they want
to signal in the capabilities (supported by platform filters, enabled by
platform filters, recognizable by BGP types).

> 2) Do you think we should do this with Flow Specification v2?
> Or should we do this instead of Flow Specification v2?

We should stop changing Flowspec v1 and shift all this new work to Flowspec
v2. It is just one new SAFI. We could perhaps also rename it to
Confspec instead of mixing it with Flowspec as number of requested new
changes or additions do not even describe the IP packet flows.

IMO there is no need to do interim solutions. For one we will never be done
with interim and it will be much harder to move to new clean solution.

> 3) Will this help operational networks?

It could help, but only in some deployment scenarios and with obvious
danger to break flowspec v1 original deployment use case.

Many thx,
Robert.




On Fri, Apr 9, 2021 at 5:45 PM Jeffrey Haas <[e-mail]> wrote:

> Robert,
>
> On Fri, Apr 09, 2021 at 05:34:54PM +0200, Robert Raszuk wrote:
> > IDR WG and AD made a decision many months back that no new features you
> may
> > be referring to sitting in your backlog will be added to flowspec v1. Did
> > you miss that - I think so based on your replies.
>
> Considering that the recommendation was based in part on my commentary on
> the work for RFC 8955 and the problems with incremental deployment of new
> features... why yes, I'm familiar with the motivations for the current
> status quo.
>
> The motivation for this draft is to see if we have a way forward in the
> interim.
>
> > > The point of the proposal is to offer an option that addresses
> incremental
> > > deployment of new flowspec features without requiring a flowspec v2.
> >
> > And that is the key src of my objection as I see what is being cooked
> here.
> > It is against already reached WG consensus. New features should go to
> FSv2
> > leaving FSv1 to do its original job.
>
> I'd request you go back to Sue's simple 3 question adoption request and
> just
> answer the questions there.
>
> Thanks.
>
> -- Jeff
>