[ppml] Re: ASN-based prefixes for leaf ASes

[Stephen Sprunk]

Thus spake "william(at)elan.net" <william at elan.net>
> On Mon, 22 Mar 2004, Stephen Sprunk wrote:
> > My proposal was that automatic /48 allocations be made to anyone with an
> > ASN, based on the premise that every AS will advertise at least one PI
> > block.
>
> If I remember right /48 is not a PI block (or rather it may not be within
> ARIN region which uses different terminology), /48 are expected to be
> allocated by ISPs (LIRs) to customers with their own network (rather when
> its individual device and then its /64) but its  provider-dependent
> (i.e. PA) as far as its being assigned by ISP where as PI would be
> micro-allocation directly by RIR

/48 is the size ISPs are supposed to hand out to leaf sites (with or without
an ASN), so that's the logical size of direct allocations to leaf sites.
This also allows encoding a 32-bit ASN in the prefix while only consuming a
/16 overall.  I say 32-bit ASN because it's clear the 16-bit space will run
out before IPv6 is retired; this could be changed to a 16-bit ASN if we
really want to assign a /32 to each leaf AS, but that seems like overkill.

> Now as far as assigning /48 for every ASN, lets not do it just yet - some
> ASNs may need larger blocks then just /48

It's hard to envision a leaf AS needing more than 64k subnets, but in that
case they could request a "normal" allocation from their RIR.

S

Stephen Sprunk        "Stupid people surround themselves with smart
CCIE #3723           people.  Smart people surround themselves with
K5SSS         smart people who disagree with them."  --Aaron Sorkin