more comments...

>Our main argument against preferences on outgoing policies has always
>been that this is the ONLY bit of policy information in the whole
>object that you CANNOT control. You can specify whatever you want, it
>completely depends on your neighbor to do the right thing. You have no
>control, your neighbor has. Besides that, the metric is *very*
>protocol dependent. In the case of BGP-4, there should be at least two
>metrics. Which is the one specified here? The definition in my mind is
>too vague, AND it is the only bit of policy that you actually cannot
>enforce.

There are cases that people do not config a preference when accepting certain routes
and rather they use the MED (Multi Exit Distriminator) information passed along with
the routes.  So the sending AS actually really specify the preference not the
receiving AS.  Let interas-out to specify its MED is to inform the receiving AS 
the policy it has and how the traffic will flow.  There is no other way except a 
phone call to let the receiving AS know.  One of the purpose of routing registry 
is to reduce the need to have phone such calls.

The MED is not just bgp specific.  IDRP has allow to specify that too.  So it
is not just for BGP.  

>I do not agree. The NACR does not specify their as-out, it specifies
>AS690's as-in. In any peer-peer agreement you have to communicate the
>desired policy to get it to work. AS690 enforces the ordering of
>announcements, not the regional. This is the whole point.

I am AS237, I want NSFNet to accept net 35 at router-eastcost as primary and
router-westcoast as secondary.  What do I do if I want to use this new syntax?  
I will say:

	interas-out: <router-east-coast> to AS690 1 announce {35}
	interas-out: <router-west-coast> to AS690 2 announce {35}

How do I express that if I can not specify MED.  I do not suppose I can write
as-in for AS690.

So as you can see, there are usage of being allow to specify MED in interas-out
especially interas-out/interas-in is really for the policy exchange between the
neighbor ASs.  I was not in the meeting of various discussion at Amesterdam but
this is how it relayed to me.

							--Jessica 



-------- Logged at  Sat May 28 20:19:21 MET DST 1994  ---------