[ppml] 2006-7 IPV6 Initial Allocation suggested changes- InputRequested

[Azinger, Marla]

Thank you Geoff.  Good points!

Marla

-----Original Message-----
From: Geoff Huston [mailto:gih at apnic.net]
Sent: Tuesday, November 14, 2006 9:31 AM
To: Azinger, Marla; ppml at arin.net
Subject: Re: [ppml] 2006-7 IPV6 Initial Allocation suggested changes-
InputRequested


But just because these numbers don't show up in your routing table 
does not mean that they don't show up in all routing tables. If I 
were to peer locally with another ISP here in Australia, then its not 
likely that this additional peering would appear in any routing table 
outside of this particular continent. So, from your perspective it 
would appear that my AS number is singly homed and it looks like 
"bloat" or "waste" to you, while to me it is necessary to undertake 
my chosen form of interconnection. So my AS, if it appears to you as 
a single homed AS, is in fact a "single transit homed" AS from your 
perspective, but that says nothing about any richer form of local 
interconnection that is not directly related to transit services.

* end of substantive comment - ramblings follow - hit delete now!

** and if you read it anyway and want to post a followup then please 
keep the followup off the ppml mailing list, as its not really 
relevant to that list - just send to me directly!

Wandering further into this area of the role of ASNs for routing , 
just for a second, the semantics of an ASN has changed over time - As 
I understand it, the original model of some decades ago equated an AS 
to both an interior routing domain and the coupling of this domain to 
an origination of a collection of prefixes that shared the same routing policy.

Over the years the "shared the same routing policy" has disappeared, 
and now the prefix is the element of routing policy, while the AS is, 
in effect, the labelling of an interior routing domain.

If we ever want to use these tokens for more than what they are today 
(today, strictly speaking, ASNs form the BGP path metric of last 
resort and the protocol's loop detector, and their use in the area of 
routing policy expression is an adornment of the original semantic 
intent rather than a core feature) and, specifically if we want to 
use ASNs as forwarding tokens, then, yes the ASN distribution 
policies would need to accommodate the aspect of one routing domain 
wanting to express multiple external routing policies using, in such 
a case, multiple AS numbers. (was that really one sentence? I think 
less coffee is called for! :-))

Its not bloat, nor is it waste. Its just numbers with meanings!

   Geoff

At 03:30 AM 15/11/2006, Azinger, Marla wrote:
>Bloat may be a bad word to use.  How about 'waste'?