Re: [Idr] IDR interim 25 September, -car type 2 - split to separate document?

[Robert Raszuk <[e-mail]>]

Hi Jeff,

If there's not always a LCM, then it's not really a route with color.  It's
> just something that's using convenient encoding - which is potentially
> reasonable.  But it's mixing the use case.
>

Well let's look at this from end to end network point of view.

I have a IPv6 based service which selectively needs to differentiate a few
infrastructure routes (aka locators or next-hops).

Maybe 5% of them.

So I may advertise my 95% of locators in type 2 CAR SAFI without any color
neither in NLRI nor in ext community. Then for those 5% I paint them with
ext community carrying some color.

My services may be available over both "colors" and depending on the SLA
forwarded to take grey (no color) or gold paths.

As you recall, the ability to aggregate routes is a very important feature
in any design. And the less we carry in the protocol the better.

Then last to me Type 2 is the proposal which addresses many previous
attempts to define separate next-hop SAFI (Ilya and I had one such attempt:
https://datatracker.ietf.org/doc/html/draft-ietf-idr-bgp-nh-cost-01)  and
as I recall we were not alone in trying to define it.

As DJ reiterated today the entire idea behind CAR is to have easily
extensible SAFI to carry various infrastructure routes. I am not sure what
value is in splitting the document now other then just to delay the
progress. This is specifically considering experimental nature of the draft
where the bar by design is not as high as for standards track docs.

Kind regards,
Robert