[ppml] FW: Policy Proposal 2007-15: Authentication ofLegacyResources

[Dean Anderson]

Thanks, Rebecca.

However, none of these reports have actual raw data from authoritative
sources such as ARIN or IANA.  They may have been generated from ARIN
raw data at some point, but the source of this data is not clear.

I have examined the pages from potaroo.net, and the source code.  
However, the source code just downloads datafiles from potaroo.net and
processes them into graphs.  Data from downloaded from potaroo.net is
not the official raw data---it may indeed be a copy, but we can't tell
unless we compare it with the official raw data.  I note also that Mr.  
Huston (of APNIC, IETF GROW WG, Potaroo.net) has previously been
involved in playing "hardball" and silencing the critics of unsupported
scientific conclusions, and Huston (through APNIC)  was the very first
customer of Vixie's DNS Root Anycast service.  I've found it best to
check on the facts asserted, and I've previously found false assertions
of fact, and claims that aren't supported by the facts.

I note that no one has so far identified a source URL for the raw data
that these reports are based on. It would be helpful if someone could
identify where this may be found.

That potaroo.net report may also be the source of the proposal to
'authenticate Legacy resources'. That report includes sections on
unadvertised space, and includes a projection based on "reclaiming" this
unadvertised space and re-delegating it to others. I believe that there
is a possiblity that the projection is based on an assumption that
unadvertised space is constant or permanent, when I'd expect it to be
constantly changing.  Sort of like the number of trucks/airplanes in for
service being counted as "unused", and assuming that they could be
"used" somehow.  This requires some work to investigate (and official
raw data), but is worth looking into.

This 'unadvertised space' may not be recoverable for a variety of
reasons. One false assumption that I noticed right away was that the
report cites "the latest BGP table", as though there were a single
canonical BGP table.  There is no such beast. Blocks may not appear in
the table Mr. Huston uses because, as someone already pointed out, the
blocks may be used for private peering or other ways that do not include
the BGP table or may not make it into Mr. Huston's copy. BGP feeds are
commonly filtered, and can be filtered at each BGP router hop.  Another
example of unadvertised space is datacenters that close and then reopen;
expansion, remodeling, relocation. Examples are endless.  During the
downtime, the IP addresses are not advertised into BGP.



I think that rationing should occur without the assumption of recovery
of unadvertised space, until any unused space is actually recovered.
After actual recovery, the rationing algorithm can naturally consider
the space in the available pool.

Incidentally, contrary to Mr. Bonomi's hyperbolic claims, all of these
reports _do_ show accelerating rates of IP Address usage.  First
semester calculus should be enough to know that an plot curving upwards
has a positive first derivative. The first derivative with respect to
time is commonly called the rate of acceleration when positive.

Also, contrary to Mr. Bonomi's claims, "decreasing exponential" is not a
term I made up and is "not like something a spammer would say".  
However, by 'decreasing exponential' I mean a function in the family of
e^(-x).

To give an example of rationing on such a function, consider that you
have 100 widgets and you want them to last 10 years, for as long as
possible.

Year 1: 100 total, give out 10 widgets
Year 2:  90 total, give out  9
Year 3:  81 total, give out  8
year 4:  73 total, give out  7

etc.  

One gives out the 1 year ration of widgets to the most 'worthy'.  
'Worthy' can be determined in a variety of ways: by those willing to
wait, those willing to pay more, and/or those with the most need.  Etc.
The selection criteria can be argued separately.

However, I do want to make this clear: 

* Both reports agree on this:  If we continue the current policy, we
will run out of space in March 2010.

* Everyone seems to agree that depletion will be a very bad event.

* It is therefore imperative to begin rationing to slow down the rate of
new delegations to conserve the available address space.

* It is necessary to do this now. One can't start rationing after the 
resources run out.



		--Dean
	

> 

-- 
Av8 Internet   Prepared to pay a premium for better service?
www.av8.net         faster, more reliable, better service
617 344 9000