[AFRINIC-rpd] Academic IPv4 Allocation Policy Second Draft (AFPUB-2013-GEN-001-DRAFT-02)

[Guy Antony Halse]

On Thu 2013-01-17 (08:56), Sunday Folayan wrote:
> Once we start referring to National laws, we will make it complex again. 

Okay, so leave that bit out.  My 3.1.2.3 covers it implicitly anyway.

> IMHO ... the allowance of 5 IPs per user more than compensates for the 
> 0.5 multiplier for part-time students. In most institutions, campus 

Again with the part-time students.  What about the part-time, contact
students who are living in our transit accomodation?  The multiplier should
be 1 for them...

The point of my formula is that it doesn't assume a static ratio that
applies universally; instead it calculates that ratio based on the realities
of the particular institution.  I expect that values will vary widely in the
range 0 to 1, depending on the particular situation of a particular HE
institution.

The way it's drafted at the moment, the only way I can fix the (frankly
incorrect) constant is by fudging my requirements for IPs-per-user.

> academic season.  The closer we are to a scenario that works across 
> Africa without hostmasters iterating and asking for more documentations 
> that could be hard for people to get, the better we are.

My formula has three inputs.  I'd be surprised if there's a HE institution
anywhere in the world that can't at the very least estimate those values to
a reasonable degree of accuracy.  But lets look at the burden of proof:

registered students - any evidence of enrolment numbers

teaching days - a term schedule

average contact days per student - in smaller and private HE institutions,
this wil simply be the same as teaching days, and easy to justify as such
(no distance students).  In larger institutions it can be calculated from
the ratio of contact to distance students.  This is simply enrolment
statistics.  We publish ours on the Internet ;-)

> Lets keep the math and definitions very simple.

I did ;-)

My formula is a statistical approximation.  It isn't necessarily completely
accurate, but it is substantially more accurate than the original one.

I could make it more accurate, at the expense of simplicity.

IMHO the improved accuracy (and resulting slight complexity increase)
actually *reduces* the risk of arguments.  Because I don't have to fudge
other numbers (specifically IPs-per-user) in the policy to arrive at a
reasonable approximation.

Put another way, I'm less likely to win an argument with the hostmasters
about IPs-per-user than I am about my university's enrolment statistics. 
See under bikeshedding[1].

- Guy
-- 
Manager: Systems, IT Division, Rhodes University, Grahamstown, South Africa
Email: G.Halse at ru.ac.za   Web: http://mombe.org/   IRC: rm-rf at irc.atrum.org
*** ANSI Standard Disclaimer ***                                    J.A.P.H

[1] http://bikeshed.com/, or
http://www.freebsd.org/doc/en_US.ISO8859-1/books/faq/misc.html#bikeshed-painting