Re: [Idr] Mirja Kühlewind's No Objection on draft-ietf-idr-ix-bgp-route-server-11: (with COMMENT)

[Marco Marzetti <[e-mail]>]

On 2016-06-14 06:28, Randy Bush wrote:
>> Let's observe that nothing prevents such "one" to extend it's inbound
>> AS_PATH for nets he learns from RS with additional AS to make it less
>> preffered ... without any special action to be taken by RS itself.
>>>> i think what might be said is "creating an undesirably longer 
>>>> AS_PATH"
>>>> and thus biasing receivers against using the exchange.
>>> 
>>> my bad (trans-pac and not enough coffee).  we're discussing the
>>> opposite, why one might want the RS to have an AS in my path.  well, 
>>> to
>>> me it would be because i have one a path directly from peer and one 
>>> via
>>> the RS.  i would prefer the former.  see draft-ietf-idr-rs-bfd-01.txt
>>> for why i kinda like to know layer two is up.
> 
> and the sender can prepend when sending to the RS.  and and and.  the
> point is not to explore all possible hacks, but to explain why this
> SHOULD is not a MUST.
> 
> life can be simple.

I raised the same question a few weeks ago on this list.
The answer was: because you can never know and IDR should always address 
all the possible cases.

My opinion is that this brings more problems than benefits (because 
stupid people do stupid things), but i think that the new lines make it 
clear enough that adding the ASn of the RS in the AS_PATH is a very bad 
idea.

-- 
Marco