Re: [Idr] the first field

[Joe Provo <[e-mail]>]

On Mon, Sep 19, 2016 at 07:40:56PM +0200, Gert Doering wrote:
> Hi,
> 
> On Mon, Sep 19, 2016 at 09:20:44AM -0700, Randy Bush wrote:
> > clue bat please.  i am making the assumption that, as we consider using
> > communities to cause remote (further than next AS) semantics, the intent

One should never assume that actions requested further than the flow 
of ones payments will be honoured. That doesn't bar the request nor 
anyone supporting said request for free, but this is my first reading 
of action at a distance. I'll go re-read the threads.
nip]
>  <the AS that should do something>:<the AS peering that is affected>:<what>

That's close to my previous "<actor>:<target>:<action>". For action at
a distnace, the onus is on the actor as to from whom they'll honour 
requests for <action> toward <target>.  Given we're positing a world
involving action at a distance, <target> should not be assumed to be
a direct adjacency. 

It would not be unimaginable for a transit provider to offer BGP 
customers the ability to signal actions for prefixes facing arbitrary
ASNs, where the customer gets the luxury of not having to monitor the
provider's adjecencies ("oh no - I have been leaking prefixes because 
they now use a third-party to reach FOO!") and the provider could simply
act locally or also tag the request for appropriate remote action.

Cheers!

Joe

-- 
        RSUC / GweepNet / Spunk / FnB / CotSG / Usenix / NANOG