On Feb 25, 2011, at 4:26 PM, Russ White wrote:
>>> You're apparently trying to separate the idea of proving an update
>>> traveled a specific path from the idea of using this information to
>>> actually filter or determine any other policy. I don't see how you can
>>> separate the two --can you explain how you can?
>>
>> If you can establish the former (that the route did travel across the path
>> it said it did), that outcome can be used as input to policy. This is
>> exactly analogous to draft-ietf-sidr-pfx-validate-01, BTW.
>>
>> Divide, conquer. A commonly used trick in computing. :-)
>
> Hmmm...
>
> This is what I thought.
>
> 1. Shane and I are arguing that you can't --or shouldn't-- imply policy
> based on which AS a particular piece of routing information went through
> multiple hops from you (I hope that's right, Shane).
"Can't"? Hm. I'll address "shouldn't", by saying if you think that shouldn't
be done, then don't configure your policy to do it. Others disagree. In any
case, the question of which AS a particular piece of routing information went
through is only tangentially related to the topic, which I will remind you is
* Is the AS-Path represented in the route the same as the path
through which the route update traveled
This doesn't speak to any particular AS in the path. It just asks whether
there is a custodial chain that extends back to the route's origin.
I'll also note in passing that as others -- Ruediger maybe? -- have noted, AS
path regexes wouldn't exist if policies weren't already being written based on
exactly "which AS a particular piece of routing information went through
multiple hops from you".
> 2. You answered:
>
>> I interpret the proposed charter item to be asking not "if AS_B *should* in
>> fact be announcing which of AS_A's routes or in what form" but rather
>> roughly "if AS_B *did* in fact announce which of AS_A's routes and in what
>> form".
>
> 3. My question was --it seems you're trying to say, "no, I don't want to
> infer policy from this, I just want to prove what happened." At which
> point I asked --what's the difference if, in fact, you intend to act
> based on what you perceive as "what happened?"
>
> In trying to "divide and conquer," you're actually assuming step 2 from
> step 1 --and the assumption is what I think we need to drive out into
> the open air and discuss by building a set of requirements other than,
> "I want to prove what happened."
You lost me again.
My best guess based on the above is that you think we should actively prevent
people from writing policies to prefer routes that have a known custodial chain
that extends back to the origin to ones that do not have such. I do not think
we should prevent them from doing so. The proposed charter addition suggests
adding a tool to enable that choice. It seems like a worthy thing to me.
By and large I think the discussion of general policy in this context muddies
the waters so I am going to try to resist the temptation to discuss it any
further.
--John
_______________________________________________
sidr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sidr
