Re: [irs-discuss] [forces] New Non-WG Mailing List: irs-discuss -- Interface to The Internet Routing System (IRS)

[Jamal Hadi Salim <[e-mail]>]

On Fri, Aug 24, 2012 at 12:24 PM, Benson Schliesser
<[e-mail]> wrote:

> At a high level, I think there is a different assumption about where the
> control-plane resides.

Clarification on the generic statement above below.

>In IRS the forwarding and control planes might be
> co-resident in the router (i.e. not "separated"), and the IRS protocol would
> be a way of extending limited control over routing state to external
> applications.
>
> If I understand ForCES correctly (which is possibly not the case) then I
> imagine it could be used in a similar architecture and/or as a protocol
> layer for IRS. But the first step in the context of IRS is to make a problem
> statement, understand requirements, etc.
>

Indeed. Once clarity sets in on the requirements, ForCES may or may not be a
good choice.

Where i feel obligated to respond is to the comment on your first part.
It boxes ForCES into only being useful in a southbound control-datapath
separation.  We've used it to control a bunch of ethernet switches which have
co-resident  control/data plane from the vendor. We've used it to configure
things that are not packet processing entities - consider this presentation
https://www.ietf.org/proceedings/84/slides/slides-84-forces-2.pdf
which illustrates how we model a configuration file (nothing to do with a
datapath functional block) into an LFB and then proceed to query/set
from the CE.

cheers,
jamal