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

[AFRINIC-rpd] Latest version of the policy AFPUB-2013-GEN-001-DRAFT-03

Andrew Alston alston.networks at
Tue Jun 25 09:32:29 UTC 2013

Hi Alan,

Firstly, there is a concern that there are institutions that will not have
their infrastructure ready in time to get space as the allocation rates
increase and the land grabbing of IP space by foreign entities becomes
more predominant.  AfriNIC themselves have admitted that they are already
seeing attempts at this, and following discussions on various
international forums, this is a topic of heavy discussion already, how to
get space outta Africa.   This policy allows for institutions to get the
space and deploy it as their infrastructure becomes available, thereby
keeping the space on the continent, generating revenue for AfriNIC and
making sure that they don't end up left behind with no space because by
the time they are ready to use it there is nothing left because it's all
being used in Asia through dodgy means.  (Again, I stress this, there is
currently NOTHING in any policy other than the soft landing policy that
prevents an organisation registered in Africa for applying for space to
use off continent, the only references that I can find to such a thing are
in the soft landing policy which will not take effect until the last /8)


One of the MAJOR concerns is how you go about justifying user concurrency
on the WIFI networks where very often the vast majority of the space is
used, and this has been a HUGE debate when getting space out of AfriNIC.

While there are various examples of this, I will refer to one specific
case in particular.

An institution provided proof of a tender that they were going on to buy
1200 Access Points to roll a campus wifi network.  Full details of this
were supplied.

The institution in question had a vast number of students and stuff, I
forget the exact numbers.  However, based on a 2:1 concurrency + address
space for current infrastructure and wired ports, the application easily
filled a /15 worth of space.

The institution in question debated this with AfriNIC for at least 6
weeks, and I'd have to go back and get the exact time frames.  AfriNIC
however refused to accept the concurrency arguments on the WIFI networks
and also refused to accept the tender notices that had been put out
publicly or the notice of intention to purchase which was sent to the
suppliers.  Their argument was that such tender notices and the notice of
intention could be withdrawn.

After a huge amount of debate, the institution in question offered AfriNIC
a compromise, give us a /16 right now if you don't believe us, and if
within 6 months we can demonstrate that the space is 100% utilised, give
us a second /16 under the first application without additional fees.
AfriNIC refused this compromise claiming that they had to close the ticket
the moment space was allocated and this would not work.

In the end a compromise was struck and the institution received a single
/16 and a /19, which today is 100% utilised and the campus is actually
gone back to using network address translation in areas to avoid having to
apply for additional space at still further expense.

Similar fights were had on 3 other applications that I am well aware of,
and despite weeks of consultation and escalating this to the highest
levels in Afrinic, nothing was forthcoming.

Education has been attempted and failed, negotiation has been attempted,
and failed, and if someone here can suggest a sane way in which to
evaluate wifi network concurrent utilisation other than by tying it to on
campus population, I am open to hearing it, but as of yet, I have not
heard any way to actually sort this out other than what is proposed, using
a multiplier.

Thirdly,  I believe that before we can really debate this further, this
should follow process and the last call should be opened as per the



>6. If the AFRINIC staff reject network plans, then it should usually
>    be for one of these reasons:
>      a) the plan has insufficient detail;
>      b) the plan does not support the number of addresses in the
>         application;
>      c) the plan appears to be fraudulent.
>    It should not be for these reasons:
>      d) AFRINIC staff demand an unrealistic level of detail;
>      e) AFRINIC staff do not understand the plan, or do not understand
>         the technology used in the network.
>    The introduction to proposal AFPUB-2013-GEN-001 suggests that
>    (d) and (e) are being used as reasons to reject applications,
>    and if this is the case then I think that should be addressed by
>    changes in internal AFRINIC procedures, and by staff training,
>    not by changes in policy.
>--apb (Alan Barrett)
>rpd mailing list
>rpd at

More information about the RPD mailing list