[ppml] Policy Proposal: Decreasing Exponential Rationingof IPv4 IP Addresses
dean at av8.com
Fri Sep 7 21:08:04 EDT 2007
On Wed, 5 Sep 2007 briand at ca.afilias.info wrote:
> Taking the monthly data, and doing sliding window averages, trends
> show up pretty quickly. There's annual cyclical variances, but year
> over year, each month should show the same trend.
> Sliding average data on 4-month window:
> Two observations:
> - monthly averages > 14M/month
> - trending upwards (however slightly)
There are some other ways to analyze the data: Compare the
month-to-month and quarter-to-quarter amounts; Keep a running sum of
the differences, and I think you'll also see that usage rate is
increasing. But, the point is: you do see the trend--there is a trend.
> Bingo. Monotonically increasing usage, and increasing rate of usage.
> We will run out, and reasonably soon. Even if the rate flattens, e.g
> at 20M/month, we will still exhaust the pool in << 5 years. Whether it
> is 3, 4, or 5 years, that's still a very short time to be fully
> IPv6+IPv4, for those who want to be around after IPv4 addresses are
> very hard to get.
I've been looking at a few old books that mentioned Address Depletion,
and IPv6. "Routing in the Internet", by Christian Huitema, Prentice
Hall 1995, and "IPng Internet Protocol Next Generation" by Bradner and
Mankin, Addison-Wesley 1996. Both books give some predictions for
depletion between 2008 and 2018. With 2007 hindsight, it looks like
Frank Solensky's use of the logistical distribution was best. In 1995,
Solensky predicted depletion in 2008. The logistic distribution, had
been used to predict growth of the telephone network in the 1960s.
Also, Solensky's prediction was based on mostly pre-CIDR data (CIDR was
introduced in 1994--RFC 1519 date is November 1993)
I'll try to resubmit my proposal on rationing again by Monday, as a
modification to the policy manual.
Av8 Internet Prepared to pay a premium for better service?
www.av8.net faster, more reliable, better service
617 344 9000
More information about the ARIN-PPML