[AfriNIC-rpd] 15-Day Last Call Period: IPv4 Softlanding Policy Proposal

[ALAIN AINA]

On Jun 26, 2010, at 7:36 AM, Mark Elkins wrote:

> On Fri, 2010-06-25 at 15:27 -0700, Scott Leibrand wrote:
>> On Fri 6/25/2010 2:49 PM, Graham Beneke wrote:
>>> This policy apparently reached consensus during the face-to-face 
>>> meeting. I was unfortunately unable to attend and I still have a concern:
>> 
>> Sorry you couldn't make it.  It was a very good meeting overall, IMO.
> 
> Agreed...
> Though I do think more time could be spent on the Policy section of the
> meeting.
> 
>>> On 25/06/2010 20:16, Vincent Ngundi wrote:
>>>> a) The minimum allocation or assignment size for IPv4 will be a /24
>>>> block (256 addresses). The maximum alocation or assignment size will be
>>>> a /23 block (512 addresses). No LIR or End User will receive more than 4
>>>> allocations or assignments during the Exhaustion Phase.
>>> 
>>> I have previously provided a set of calculations and proposed a 
>>> maximum allocation limit that may far more appropriate.
>> 
>> This was discussed at one point during the policy discussion in Kigali, 
>> FWIW.  Others may remember better than I do the specifics of that 
>> discussion, but...
>> 
>>> I would be interested to see the resource planning calculation that 
>>> allowed us to arrive at a limit of /23.
>>> 
>>> Limiting allocations to such a tiny value makes applying for IPv4 
>>> space virtually pointless. We might as well reserve the whole /8 and 
>>> not just the last /12. It will have the same effect...
>> 
>> I don't think the entire /8 would be tied up and unusable.  As best I 
>> can tell, this policy, if implemented as written, will have the result 
>> of rationing medium and large ISPs from continuing to acquire new IPv4 
>> space for assignment to their customers after they've received four 
>> /23s, which in turn will encourage those ISPs' larger customers to go to 
>> AfriNIC directly to get space under end user policies.  This will result 
>> in more organizations interacting with AfriNIC directly, and less 
>> aggregation within the last /8 than would occur without this policy.
> 
> It was pointed out - if you take the total number of members as being
> about 1000 (www.afrinic.net/statistics/member_stats.htm) and gave each
> member four /23's - there would still be lots of space left.
> 
> Then again - this policy will probably only be in effect in about 3
> years time (minimum 2 years),
> 
> It may make sense before then to re-visit the policy and either:-
> 
> 1 - Change the '/23' to something like a '/22' (or '/21') - leaving
> everything else as is....
> 
> -and/or-
> 
> 2 - Change the '/8' policy to only kick in when - using the existing
> policies - either an allocation or assignment will touch the last
> '/10' (or '/11')  of (contiguous?) space left of this last '/8', deny
> that request and proceed with the policy as is.
> 
> ...but the fundamentals are pretty much fine.


 I don 't reasons to rush to adopt  this policy  and revisit it later.

 I suggest that we prolong the discussions  and investigate more for the adequate  scenario.  I am seeing something similar to mark's  second  point  in the following format:


-Phase 1 : Status Quo

We continue with the current policy until  we reach /x

-Phase 2: Allocation/Assignment for transition to IPv6

We use the policy as it with:

a minimum of     /y 
a  maximum  of  /z

With reserve of /12  

We will look at the stats and trends and get some values for X,Y,Z


 Hope this helps

--alain