[ppml] HD ratio proposal vs. Changing percentage burden

[Bill Darte]

Michael Dillon wrote...
> 
> Policy Proposal 2004-2 proposes that we change the 80% threshold 
> for additional IPv4 address allocations to a sliding scale 
> formula based on the same type of Host Density (HD) Ratio as 
> we currently use for IPv6 allocations. The formal policy proposal 
> is on the ARIN site http://www.arin.net/policy/2004_2.html
> and is up for a vote at the Vancouver meeting.
> 
> This is a significant change to the way that IPv4 allocation
> is done and I have prepared some additional background
> documents to explain and clarify the proposal. If you go
> to the ARIN page mentioned above, at the bottom there is
> a link to the background documents page. There you can
> find a brief slide presentation, a spreadsheet, some PERL
> scripts and some tables. Please take a few moments to look
> through this material. If there is anything unclear about
> the proposal, I will try my best to answer questions on
> this list.

Reading your paragraph above, I was (again) struck by the contrast
between the complexity of HD Ratio allocation and its utility.

"A slide presentation, a spreadsheet, some PERL scripts and some tables"...
and "if anything is unclear...I will try my best to answer questions"

Seems to me that most people who must deal with the IP provisioning
(allocation & assignment) process are looking for administrative ease and
fairness in such policy.

HD Ratio trades administrative ease off for fairness. Using HD Ratio
increases 
the efficiencies (fairness) of provisioning policy by providing a more
accurate
assessment of utilization.  This applies to all users, but its real value
(relative to the 80% rule)is when large blocks are provisioned finely
through a more extensive hierarchy. Not every segment of the industry
'truly' benefits but everyone would be subject to the increase
complexity...all this as I understand it.

Another way of achieving the same utility may be to change the 'percent
utilization burden' for various entities with more or less extended
provisioning hierarchy.  You have a small hierarchy therefore you live with
80% (current standard)..others with increased hierarchy suffer a burden of
only 70% and still others with the most extensive hierarchy use 60% as their
threshold.  Is there a way to measure the extensiveness of
hierarchy...simply?  This process would presume that you could conveniently
(simply) measure or fairly assume hierarchy...and one would have to overcome
the perception that organizations were being unfairly treated differently.
The latter issues is really the same with HD Ratios, but is masked by its
mathematical elegance and complexity...IMO.

Both cases for change require an educational process and acceptance by the
'majority' of those active in ARIN policy matters in order to overcome the
inertia of the status quo.

Can hierarchy be measured simply?...or can it be assumed fairly?  Seems that
all else related to HD Ration flows from this question.  Paul Wilson
provides a chart
http://www.apnic.net/mailing-lists/sig-policy/archive/2003/08/msg00000.html
that "proposes a set of assumed hierarchical depths which may be reasonably
expected within
hierarchically-managed address spaces of certain sizes."  Is there clear and
concise empirical evidence to support these assumptions?  

Also, could you comment on why HD Ratio would be the preferred approach to
change? Is the educational burden less to overcome the status quo of
administrative ease than simply changing the percent utilization burden?

I have no preference (officially), I'm just a little confused by these
alternatives to change.

Bill Darte
ARIN AC