[governance] IPv4 allocation (US Gov version)

[McTim]

Hi Loius, 

Let me try to expalin again:

On 9/3/05, Louis Pouzin <pouzin at well.com> wrote:
<snip>
> 
> On Fri, 02 Sep 2005 09:43:29 -0400, Milton Mueller wrote:
> 
> >The USG comment's method of accounting probably does not count the allocations  made prior to the creation of the RIRs. In other words it excludes the prior allocation of around 80(? - just a guess, don't have access to the information here) /8's and concentrates only on what has happened since, say 1999 or so.
> 
> Possible. Actually I am not trying to explain out the US Gov arithmetics, it's their job. Rather I am interested in the factual "cumulative IPv4 address allocations", as they state.

I think Milton's point is that you cannot reliably ascertain what you
wish to know from this page.

> 
> On Thu, 1 Sep 2005 19:27:47 +0300, McTim wrote:
> 
> >ARIN up until recently allocated and assigned resources to all of
> Latin/South America/Carib and Africa south of the Equator.
> 
> >Up until quite recently the RIPE NCC allocated and assigned resources to all of Africa nouth of the Equator.
> 
> >Perhaps this is where your calculations differ.
> 
> LACNIC got its first block in Nov 02, and ARIN got one block in Aug 02. ARIN's allocation did not change. LACNIC got a second block in Apr 03. I don't have older tables to see if this block was transferred from ARIN or taken off unallocated space, as seems to be common practice for extensions.
> 
> AFRINIC got its only block in Apr 05. It came from reserved space without altering RIPE's allocation.
> 
> Thus, those new RIR's increased by a mere 3% the total amount of allocated blocks, without changing significantly the geographical distribution pattern. This cannot account for the 40% gap with the US Gov figures.

I think you missed the point.  RIR databases hold records for some of these /8s.

ARIN Database holds records for much of this pre-RIR space.
The RIRs call this "legacy space" or "Early Registration".  An Early
Registration Transfer project has been ongoing to try and get the
space in the Db where the space is actually in use.  So EU bodies that
got InterNIC space have been transferred tot eh RIPE db, etc.

> 

> 
> That's the real bone. I counted them as USA allocations, because there is no mention of RIR or any other institutions. They total 50 blocks. That's a huge lot, more than RIPE, APNIC, LACNIC, and AFRINIC lumped together (40 blocks). To me it smacks of a walled garden.

It is the early registrations. They should NOT all be counted as USA
space. This will account for at least some of the discrepancy.

> 
> I checked a sampling of 340 addresses allocated to Various Registries. They look like a random sampling of site names as found routinely in the internet. See examples further below. Even though the names may be registered under about any gTLD or ccTLD, the addresses do not belong to the RIR's allocations. Who got these addresses and how ? Tracing the routes would reveal more details on their network attachment, but that's for further study.

I don't think name tracking will be so helpful here.  Looking at the
routing table will be more difficult than looking directly into the
RIR Dbs.  Having said that, you will need to get copies of the RIR
databases and do your queries locally if you want to do comprehensive
searches for locations of address blocks.
You will need to write to each RIR and ask them for a copy of the
data, as you will be rate limited tothe number of queries you can do
via scripts.  You still won't get locatinos for 100% of the blocks.

> 
> Yes. Which addresses are used and by whom is a big black hole. But this is an entirely different issue. For the moment I am concerned with allocation, not use, and no deeper visibility than the IANA allocation.

Then perhaps the US numbers are finer grained than yours.
If you count the "Various Registries" blocks as USA space, you will
get results that do not map to reality.

> 
> Just to play with various stats methodologies I ran a few more accounting methods. x % means x % of the total of allocated blocks.
<snip>
> 
> Now, anyone can guess what is factual. 

or do the hard work to actually find smt that approximates reality.  

OR 

You can re-read the US commentt, which I have just done, I think they
are only counting allocations since the creation of the RIRs.  Here is
the quote that leads me to think this:

"Continued internationalization of the Internet is
evidenced by the recent creation of Regional Internet Registries
(RIRs) for Latin America and
Africa and the enhanced efforts of the Internet community to work
towards an equitable
distribution of IP addresses. For example, as of June 2005, cumulative
IPv4 address allocations
had the following geographical breakdown – 33% to the Asia Pacific
Region, 32% to North
America, 31% to Europe, 3% to South America and 1% to Africa. For that
same period
cumulative IPv6 allocations were – 56% to Europe, 23% to the Asia
Pacific Region, 17% to
North America, 3% to Latin American and 1% to Africa."


"For that same period" in re IPv6 leads me to think that they are only
counting recent allocations. This easily explains the discrepancy.


-- 
Cheers,

McTim
nic-hdl:      TMCG

_______________________________________________
governance mailing list
governance at lists.cpsr.org
https://ssl.cpsr.org/mailman/listinfo/governance