[arin-ppml] 2014-14 Transfer Statistics analysis
springer at inlandnet.com
Thu Oct 9 09:15:30 EDT 2014
forwarded here by kind permission of Owen DeLong
I hope that this will inform our discussions.
---------- Forwarded message ----------
Date: Mon, 6 Oct 2014 10:08:10
From: Owen DeLong <owen at delong.com>
Subject: [arin-council] Transfer Statistics analysis
Based on today?s discussion, I took the liberty of putting together some analysis. The spreadsheet is included at the bottom for those that want to look at the raw numbers. I also included a PDF of the spreadsheet for those that may not be able to read a numbers spreadsheet.
I imported the prefix lists for 8.3 and 8.4 transfers from ARIN?s website: https://www.arin.net/knowledge/statistics/transfers.html
Then I sorted by prefix size (descending prefix length) and created a column which computed the number of /24 equivalents represented by each prefix. In the next column, I summed the number of /24 equivalents covered by any policy affecting /n and longer (smaller).
The next column shows what fraction of transfers (actual number of transfers or roughly the percentage of ARIN transfer workload) is impacted by a policy covering /n and longer (smaller).
/24s represent 0.07% of the address space transferred and 20.09% of the number of transfers completed.
/23s and longer (23 and 24) represent 0.14% of the address space transferred and 29.22% of the number of transfers completed.
/22 and longer are 0.35% of address space and 43.38% of all transfers.
/21 and longer covers 0.53% of address space and 49.32% of transfers completed.
/20 and longer covers 1.29% of address space and 62.1% of transfers completed.
/19 and longer covers 2.16% of address space and 69.41% of transfers completed.
/18 and longer covers 2.92% of address space and 72.6% of transfers completed.
/17 and longer covers 4% of address space and 74.89% of transfers completed.
/16 and longer covers 17.47% of address space and 89.04% of transfers completed.
IMHO, the address space column is a good measure of risk and the percentage of transfers completed is a good measure of benefit to any policy exempting needs basis testing for /n and longer. As such, I?ve added another column which shows the reward/risk ratio computed from these two measures.
Personally, I think /20, covering 60+% of all transfers while risking roughly 1.29% of the address space is a good place to balance the risk for an initial cut at this policy. I might be persuaded with strong community support to accept /18 at 2.92% of address space and 72.6% of all transfers. Exempting almost 90% of all transfers from needs testing seems very far to go on an initial experiment.
-------------- next part --------------
ARIN Proprietary & Confidential Information
ARIN-COUNCIL mailing list
ARIN-COUNCIL at arin.net
More information about the ARIN-PPML