[ppml] Policy Proposal 2003-10: Apply the HD Ratio to All Future IPv4 Allocations

Michael.Dillon at radianz.com Michael.Dillon at radianz.com
Tue Sep 16 11:17:00 EDT 2003

>Is there a good reason to abandon the relatively simple 80% rule in favor 

>of a mathematically more complicated formula that effectively drops the % 

>utilization rules by a noticable amount?

Yes. Most of this is explained in RFC 3194
and in the APNIC paper whose URL was incorrectly
given in the original posting of my justification 
of the policy. 

In a nutshell, organizations managing larger hoards of IP addresses
cannot maintain the same level of efficiency due to the overhead
inherent in subnetting a large block into many small fragments. In
order to not penalize them for losing these "crumbs" of IP space, 
the usage ratio should be lower for larger hoards of space. The HD
ratio is a neat and simple calculation that achieves this goal.

>I'd always assumed the 80% applied to all of your IP space.  From the 
>above, it's not clear.  The first bit clearly says it applies to your 
>recent allocation.  The second bit suggests to me that it applies to all 
>prior allocations.  If it doesn't, then what's the definition of 
>"efficiently utilized all previous allocations"?

The assumption is that if you achieve 80% efficiency on a block today,
then it will be that way forever. Since the policy only requires ARIN
to examine the most recent allocation (saving on costs) this opens
a loophole that larger networks can take advantage of. When they get
a new block, they use these addresses first (LIFO inventory queue)
but if they run out before there next ARIN allocation then they still
have spare addresses left in the 20% or so of the previous allocations
plus in the space released through customer churn.

>If that's the reason for suggesting a 0.930 HD requirement on the most 
>recent allocation, why not just come out and say 50% for the most recent 

Because I think that the HD ratio is a better idea and it is simple to
calculate. I also think that it is better to evaluate an ISP's total
hoard of addresses rather than only the last one. I'd like to see the
more flexible HD ratio in use.

--Michael Dillon

More information about the ARIN-PPML mailing list