Search RPD Archives
Limit search to: Subject & Body Subject Author
Sort by:

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

Andrew Alston alston.networks at
Mon Feb 11 10:25:23 UTC 2013

Hi Douglas,

See responses inline

> All,
> Sorry I am late to this party, but I would like to voice my principle
objection to this policy because it reads more like an
operational/implementation document rather than a policy. My feeling is that
inconsistent treatment of resource applications is at the centre of this
> If I am right then I would posit that this is not the first time we see
some attempt to resolve administrative issues with policy, so for me it
would be important to put in place/fix a mechanism for AfriNIc and the
community to resolve these kind of issues.

In some sense I agree with you on this.  The inconsistent treatment of
resource applications is one of the drivers of this policy.  The other
driver is the lack of ability for an external organization to truly
understand the needs of a University in terms of these resources simply
through the examination of network documentation, since as has been
demonstrated on this list multiple times in previous weeks, network
documentation does not talk to concurrent usage, especially when it comes to
wifi systems.  Secondly, I would argue that many policies are attempting to
resolve issues.  I have always been of the strong belief that you do not fix
something that is not broken, however, if it is broken, you use what
mechanisms are in place to rectify the problems.  As such, all policies will
attempt to resolve issues, the question is how specific that issue is and
how large the impact of that issue is on a particular individual or the
community as a whole.  I would argue that this policy addresses issues
covering a large and growing segment of the AfriNIC user base and is hence
justified, rather than attempting to resolve specific issues for one
specific member.

> More specifically 3.1 of this policy:
> > 3.1) To qualify for address space, Academic institutions will need to
apply as end users and provide the following >documentation:
>>3.1.1) Proof of Institution's registration/accreditation
>> 3.1.2) Proof of the number of registered full time students
>> 3.1.3) Proof of staff head count.

> I am averse to the idea of hardcoding requirements here as it will
constrain staff from doing any extra due dilligence --- applications can
refer them to #3.1 when probed :-). imagine how abused this scripted process
can be --- we should allow some discretion to staff - 
> unless offcourse we have a problem with the way they are  doing their job,
in which case we should try and fix that, only I am doubtful that this would
fall under the purview of the PDP

I don't think it's a case of having an issue with how they do their jobs.  I
think it is more a case of if individuals at an RIR have the experience,
skills and capabilities to adequately analyze the requirements of networks
as large and as complex as these, especially when it comes to how much
resource they need to operate.  In a world before wireless infrastructure
and coverged mobility, it was easy to judge, IP's to devices.  When you
implement large complex wifi networks, concurrency becomes a major issue and
that is directly tied to the number of people on a campus and their usage of
such.  How big do you make the subnets.  When an institution is going out on
procurement of such infrastructure, they need to know they can number it and
they have to make assumptions based on their student base, and it ends up in
a catch 22.  They cannot provide AfriNIC proof of usage since they don't yet
have the equipment nor do they have the IP space to demonstrate anything,
and I would argue that at that point ONLY the institution has the
capabilities to make those judgments, since only they know the details of
their environment.   Hence, allowing for a ratio which has a maximum but
allows for people to apply for less if they believe they need less than the
required maximum

> Also 3.5:
>>3.5) Under the policy, HEI shall be eligible to receive IPv4 resources 
>>at a ratio not less than 5 IPv4 addresses per campus user, where campus 
>>user is defined in 3.2).
> It would appear that from the thread 3:1 would be something members see as
pragmatic and are able to agree to, although some numbers have been advanced
to demonstrate a  5:1 situation. My take on this particular discussion is
that I would agree with the demonstrated 3:1 that most of the people on >
the list seem to lean toward, albeit, if we accounted for the future growth
of the Internet (stats by Andrew :-)), you will agree that a 5:1 is wise as
it would support scalability in the near future. For this reason I would be
in support of a 5:1. So I feel the following would be a reasable compromise
> 1.	No minimum ratio
> 2.	> 3:1 screening begins
> 3.	5:1 is the max

I don't have issues with (1) and (2), I would raise (3) to above 5 though,
since I would say that 3 is almost the minimum that an institution not
running NAT is likely to be able to cope with if they have any form of
wireless infrastructure, and they may well need more, but I would hesitate
to place a maximum if it can be justified as higher.  In order however to
prevent problems, a maximum may be useful but I would place it at 7 or 10.



More information about the RPD mailing list