Search RPD Archives
[AFRINIC-rpd] IPv4 Address Allocation and Assignment proposal
alston.networks at gmail.com
Mon Feb 11 09:41:18 UTC 2013
I need to start this email stating that I write entirely in my personal
capacity here, and the views represented here are not those of any current
or former employers and/or clients.
As the ex CTO of TENET for 6 years, I need to disagree with what you said
below for the following reasons.
a.) There is no financial incentive to get IP space through an NREN, in
fact, the reverse is true. NREN space is automatically LIR space, the LIR
fees are vastly different to those of PI fees, and an NREN, being a
non-profit organization has to recover its costs from its member base.
Hence, if an institution chooses to apply for space via an NREN, the NREN
has to recover the LIR fees from the institution and the financial viability
disappears. The academic discounts applied by AfriNIC apply to both NREN's
and direct academic applications (as well they should), and PI space is
simply far cheaper.
b.) While NREN's are typically the primary means of access for many
universities, many universities also choose to multi-home. This cannot
happen without PI space and/or de-aggregation of the NREN space. It would
force the NREN to allow institutions that multi-home to announce subsets of
aggregated space under the institutions ASN. This multi-homing is becoming
more and more of a reality, and I have actually seen two separate audit
reports for universities where the auditors are starting to demand it (in
the same way they demand redundancy on critical IT infrastructure on the
server level). I know of institutions in both South Africa and Tanzania
that also do this form of multi-homing
c.) Institutions are looking at peering directly at IXP's primarily for
collaboration on research projects with commercial entities, where such
point to point links are mandatory, this requires non-LIR space.
d.) The provision of commodity internet access by NREN's, while common,
is not a universally accepted model. It is currently the model in place in
both KENET and TENET and most of the African NREN's, however there is no
guarantee this will continue to be the status quo. There are already
institutions that ONLY get academic access via their NREN's and get their
commodity elsewhere (I can think of a Kenyan example for this). Again, this
requires non-LIR space.
So, from the perspective of someone who has actually worked in the NREN
environment, I have to disagree with you on these points.
From: Walubengo J [mailto:jwalu at yahoo.com]
Sent: Saturday, February 09, 2013 9:35 PM
To: Owen DeLong
Cc: Andrew Alston; sm+afrinic at elandsys.com; rpd at afrinic.net
Subject: Re: [AFRINIC-rpd] IPv4 Address Allocation and Assignment proposal
Been planing to get back to you on one point. That the KENET, TENET or the
NRENs (National Research and Education Networks) model does NOT stop
individual universities from from applying for IP resources. However, there
are very strong (financial) and other incentives for doing so through the
NRENs (I believe around 50% rebate for IP resources).
Further International bandwidth is still expensive in Africa, so most
universities would benefit from the economies of scale by pooling together
their bandwidth requirements and purchasing the same in bulk from
international markets. So once you have pooled together on the bandwidth
level, it does not hurt to do the same with IP requests.
From: Owen DeLong <owen at delong.com>
To: Walubengo J <jwalu at yahoo.com>
Cc: Andrew Alston <aa at alstonnetworks.net>; "sm+afrinic at elandsys.com"
<sm+afrinic at elandsys.com>; "rpd at afrinic.net" <rpd at afrinic.net>
Sent: Friday, February 8, 2013 10:29 AM
Subject: Re: [AFRINIC-rpd] IPv4 Address Allocation and Assignment proposal
On Feb 7, 2013, at 22:50 , Walubengo J <jwalu at yahoo.com> wrote:
As one of the board members, I can assure that I do read widely the posts on
this rpd list. And as you rightly hint, I try hard not to be too liberal
with my views. I believe Board members must ride the delicate path of not
being seen as too pre-emptive, biased or exerting undue pressure on the
direction the discussions should go. But since you have drawn me out of the
woodwork, I cannot resist to share my views on this Policy - particularly
because of my affiliation with Academia.
And straight to the point of ratios. Whether it is 1:1, 1:3, 1:5 /etc IPs
per student/user in a University - this is largely applicable in SA, or
middle-income type of economies where No. of students is closely correlated
with the expected No. of PCs/Laptops/IPad/etc. In sub-saharan africa, I
can tell you that the link between No. of students and No. of PCs is largely
non-existant (though it may slightly exist in private universities - again
where income levels of student communities are higher).
Yes, but, Walu, do you consider that situation desirable, or would you like
to see it evolve to be more like you describe SA?
Is there significant harm in making the addressing available in advance to
facilitate that evolution?
I work in a public university (which in most sub-saharan countries) would
take over 80% of the total University Student enrollments in a given
country. I can tell you for sure that the the No. of PCs in a Public
University in a Sub-saharan economy has more to do with the ICT Investment
ratios rather than student numbers. In other words what % of the University
budget is going into ICT development per year - which translates better to
the No. of Labs, IP Devices expected. Perhaps you may want to review your
policy in this light.
This is the kind of thinking that I was referring to when I mentioned to Nii
that I don't see the point in using equipment poverty as a reason to
preserve address poverty. I agree that both issues need to be addressed.
AfriNIC cannot solve the equipment issue, at least not completely or
directly, but can solve the address issue. Further, I don't believe this
policy would force a university to apply under the ratio criteria.
Universities that wanted to could still apply under the existing policy and
get less space if they had some reason and desire to do so.
Another point regards the fact that in recent years, Higher Educational
Institutions have the trend of acquiring IPs through their Educational
Networks e.g. TENET in SA, KENET in Kenya, etc. This trend is being
picked up in West Africa and other parts of Africa (e.g.Madagascar). And So
in future, less and and less individual universities may be applying
directly to AfriNIC for space. This should however not stop the policy -
since the same policy would still be useful in informing the Educational
Networks on how to approximate IP resource capacities in order to
subsequently request from AfriNIC.
IMHO, that's an unfortunate trend that probably indicates that getting space
from AfriNIC directly either is, or is perceived to be more difficult than
it should be.
Finally, with the continued use of NAT - African Universities tend to have
a bigger problem of announcing their allocated space than perhaps the
problem of getting them from AfriNIC :-). I believe there is a policy
requirement that one should announce a % of their IP space within a given
time frame but not sure if this is monitored and/or if there are examples of
consequences for not doing this.
IMHO, we should discourage the use of NAT in universities where feasible.
NAT is contrary to good security practices and contrary to the principles of
a free and open internet.
On other matters that have been discussed:
*Policy vs Operational Issues: I agree we need to be careful not to use an
IP resource policy to address Operational Issues. If AfrinIC staff and
management are taking too long on processing IP resource request, there
should be channels to address this e.g. by ensuring AfriNIC reviews or
publishes its Service Charter/SLA committement to its membership.
Agreed. I think this policy proposal is intended to resolve a policy issue
where AfriNIC staff in its strict application of current policy is making
university assignments and allocations unnecessarily difficult. I don't
think it is intended to directly address timeliness or other operational
concerns. Admittedly, a simplified streamlined policy for allocations and
assignments is likely to result in an improvement in response times, but I
don't think that is a primary intent of the policy so much as a beneficial
second order effect.
*Inter RIR IPv4 Transfers:
I did agree with the message that hording IPv4 space in the face of emerging
IPv6 space is like holding onto a losing stock. A time is fast
approaching when Africa will be boasting of having millions of IPv4 space
that is not useful to anyone. But this reality should trigger us to come up
with strategies of increasing usage as opposed to strategies of "selling"
or liquidating the stock.
I think this policy is intended to create increased usage and more effective
distribution to entities that should be using the addresses (even if they
lack the financial resources to take full advantage at the moment).
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the RPD