> On Feb 27, 2015, at 01:43 , Izumi Okutani <iz...@nic.ad.jp> wrote:
> 
> On 2015/02/27 17:58, Usman Latif wrote:
>> I think organisations that have obtained portable address ranges from RIRs 
>> should have the liberty to use public ASNs from day one (if they want to) 
>> regardless of whether they are single homed or multihomed.
>> 
> 
> OK, that's an interesting approach.
> 
> What is the reason for this? Would be curious to hear from other
> operators as well, on what issues it may cause if you are a single homed
> portable assignment holder and cannot receive a global ASN.

I can see a few reasons.

1.      The difficulty of renumbering from a private ASN is proportional to the 
number of links,
        not the number of ASNs. Ergo, someone who is single homed, but plans to 
become
        multihomed at some unspecified date in the future may, indeed, have 
good reason for
        wanting to do so with a public ASN.

2.      I see very little harm in adopting such a policy, so long as it is 
limited to one ASN per 
        organization.

3.      If you have multiple links to a provider with diverse topology, it is 
desirable to be able
        to use a routing protocol in order to prevent black-holing traffic 
across down links, etc.
        The only routing protocol any sane ISP would run with an unrelated 
third party is
        BGP. BGP requires an ASN. See above for why a public ASN may be more 
desirable
        under this circumstance than a private one.

As to the references to RFC-1930, I think they are anachronistic at this point.

RFC-1930 was written before 32-bit ASNs were available and with a strong eye to 
the
coming shortage of 16-bit ASNs. While I agree that even the 32-bit pool of ASNs 
is finite,
I don’t think we’re going to cause a shortage of them by allowing single-homed 
organizations
with PI space who plan to multihome at an unspecified future time to receive 
one.

As such, I believe such a policy would do no harm and provide benefit to some 
members
of the community. If it were proposed, I would support it.

Owen

*              sig-policy:  APNIC SIG on resource management policy           *
_______________________________________________
sig-policy mailing list
sig-policy@lists.apnic.net
http://mailman.apnic.net/mailman/listinfo/sig-policy

Reply via email to