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

[Andrew Alston]

Exactly SM, 

The head count issue with the ratio is a compromise position to serve both
requirements.  The ratio in question is where the debate is.  As both Sunday
and I have stated, the original ratio was set at 3, it was the community on
this list that seemed to reject that and want it higher, so we raised it to
5.

The number 3 was based on research done across a fairly large cross segment
of campuses and was what we found to be current at this point, I will admit
it does not allow for much growth.  Interesting enough, I have an
institution that has 38 thousand students on it, they currently have a /15
worth of IP space.  That space as of this week is now completely utilized.
Furthermore, the wifi deployment on that campus is far from complete, there
are an additional couple of hundred access points still to be deployed.  So
the maths, they are already running at 4:1.  It was only in the last week
that we had to adjust the lease count for one of the PC labs there because
even though PC's that are not in use are turned off, we ran out of leases
because the labs were full, 1500 machines in there alone 100% utilized.

The 5:1 ratio allows for expansion, it allows for what is becoming a growing
trend.  Africa if I recall has one of the fastest growing rates of wireless
and mobile connectivity in the world (largely driven by lack of physical
infrastructure), with Kenya actually leading this pack.  

I am quite happy with a 4:1, or even a 3:1, it would be hellishly tight, but
at least its livable.  But at the same time, I accept the so far expressed
community view that a 5:1 is a better ratio and would more happily support
that.

Andrew
 

-----Original Message-----
From: SM [mailto:sm at resistor.net] 
Sent: Friday, January 25, 2013 11:32 PM
To: Andrew Alston
Cc: rpd at afrinic.net
Subject: RE: [AFRINIC-rpd] Academic IPv4 Allocation Policy Second Draft
(AFPUB-2013-GEN-001-DRAFT-02)

Hi Andrew,
At 12:37 25-01-2013, Andrew Alston wrote:
>Aside from which, how many times do I have to say this, *the designs 
>were submitted, the documentation was submitted, the proof of purchase 
>was submitted, the whole lot*

There are various approaches to network design.  Two networks, each serving
the same number of users, may not necessarily need the same amount of IP
addresses.  If I have to go into the details of efficient use I would have
to review the design.  The question that comes to my mind are:

   (a) Is it up to me to tell you how you must design your network?

The answer you have given is no.

   (b) Is it up to me to identify whether there is adequate justification
       for the IP addresses requested?

The answer to (b) may be yes.  However, as I do (b) I should avoid (a).  For
example, it is not for me to say that you should use NAT for part X of your
network as it's isolated.  If I understood the problem correctly it is about
finding a middle ground between adequate justification and stringent
justification, and avoiding the situation where I may be telling you how to
design your network.  This proposal suggests using head count as a measure
to make the request process easier.  The next step would be to find the
acceptable ratio.

Regards,
-sm