Implementation of RIPE-181++

Laurent Joncheray (lpj at merit.edu) on September 20:
> 	(a revised version)
> 	Just a question about the policy syntax and its implementation:
> 
> o AS1 peers with AS2 at nap1, nap2, nap3
> o AS1 accept from AS2 the networks in AS100 at nap1, nap2, nap3
> o AS1 accept from AS2 the network {10/8} at nap3 *only*
> o 10/8 is not in AS100
> 
> 	- How do you express this policy with the RIPE-81++ syntax (as-in/
> 	interas-in stuff...)
> 	- What's the algorithm used to generate the filter list (list of nets
> 	accepted) of AS1 at nap1, nap2, nap3

I think this question and the one I asked a week back are both due to
the ambiguity on the relation between as-in and interas-in. I think
the relation between as-in and interas-in needs to be made clear and
explicit.

>From the perspective of someone who read ripe181 Sep 16 version, all
solutions proposed do not solve Laurant's problem. I.e. solutions are
not consistent with ripe181 document.

Marten Terpstra (Marten.Terpstra at ripe.net) on September 20:
> aut-num:    AS1
> as-in:      AS2 100 AS100 OR {10.0.0.0/8}
> interas-in: AS2 as1nap1 as2nap1 AS100
> interas-in: AS2 as1nap2 as2nap2 AS100
> interas-in: AS2 as1nap3 as2nap3 AS100 OR {10.0.0.0/8}

ripe-81++ says pp 27-28 in ascii version:

   If the global policy mentions more routes than the local policy
   then local preferences for these routes are assumed to be equal for
   all links.

My interpretation is as follows:
	as-in:      AS2 100 AS100 OR {10.0.0.0/8}
mentions more routes than
	interas-in: AS2 as1nap1 as2nap1 AS100.
I.e. it mentions {10.0.0.0/8} extra.
Hence,
"local preferences for {10.0.0.0/8} are assumed to be equal for all
links".

All links include "as1nap1 as2nap1" Hence, AS1 accepts {10.0.0.0/8} on
nap1 as well.

Marten Terpstra (Marten.Terpstra at ripe.net) on September 20:
> aut-num:    AS1
> as-in:      AS2 100 AS100 OR {10.0.0.0/8}
> interas-in: AS2 as1nap3 as2nap3 {10.0.0.0/8}

This will make AS1 accept {10.0.0.0/8} on nap1 and nap2 as well.
Since policies on those naps will be the one in as-in.

Also interesting to see that this solution contradicts the previous
one:
	Previous solution is correct if local policies did not copy
		extra routes from global policies.
	This solution is correct (only for nap3) if it did.

Perhaps, with the current ripe181 document, there is no correct
solution. If this is the case, Laurant owes me a beer.

Cengiz

-- 
Cengiz Alaettinoglu       Information Sciences Institute
cengiz at isi.edu            University of Southern California




-------- Logged at  Tue Sep 20 21:43:14 MET DST 1994  ---------