Re: [GLOBAL-V6] Re: I-D ACTION:draft-narten-ipv6-3177bis-48boundary-00.txt

[Thomas Narten <narten@us.ibm.com>]

> On Jul 19, 2005, at 10:28 AM, Thomas Narten wrote:
> > And to be clear, I don't believe anyone is proposing this. I.e., going
> > from a /56 to a /48 would require renumbering, most likely.

> I think it'd be appropriate to do the binary split sparse allocation  
> approach, but that's an implementation decision.  Given IPv6  
> technology right now, I think it would be prudent to reduce the cost  
> of renumbering if possible.

Me thinks this may get a bit complicated.

I believe that we want the RIR policies to be such that ISPs have no
incentive one way or the other in terms of how much addresss space to
allocate to end sites. If an end site can justify an allocation of
size X, it should get it. ISPs should not be able to push back on such
requests using arguments like "giving you more costs us more in terms
of RIR costs", etc. 

But if ISPs do sparse allocations (or something similar where they
hold space in reserve) that will need to be factored into how RIRs
measure the "utilization" of an ISP's block. This will make things
more complicated. (At least some thought in this space is needed
though, since the current policies just count /48s.)

Also, it's not immediately clear to me that having ISPs keep large
chunks of (effectively unaggregatable) space in reserve is a Good
Thing. If they do so, would moving the /48 boundary to /56 really
provide the long-term savings we're aiming for?

Thomas