[arin-ppml] An article of interest to the community....

[Jimmy Hess]

On Tue, Aug 30, 2011 at 9:41 AM, Milton L Mueller <mueller at syr.edu> wrote:
> These kinds of comments show a certain lack of sensitivity to the needs of actual network operators. Given the pace of the IPv6 transition,
Reality is even worse than he stated....  because he implied there is
a way to use "IPv4 more efficiently".

I would equate this to using up all the RAM in your computer and
deciding,  your applications  will need to utilize the available
memory more efficiently  (which means swapping pages to disk),
because NAT with IPv4, and dynamic addressing,  is  basically a
similar principal to a computer system using a page file,  so the same
bit of virtually addressed physical memory can be shared between many
processes.

The problem is you don't actually increase the amount of available
resource   -- swapping  seems to work very briefly,  but then your
working set exceeds the available resource    (E.g. the number of
public addressing you need)  exceeds  the available addressing.
And there will be a breaking point beyond which the system implodes.

The IPv4 analogy to system thrashing would be...    address churn,
NAT-related internet experience degradations, and routing table
explosion with increasing numbers of smaller and smaller allocations /
resources freed up to satisfy a _fraction_ of actual requests.

More importantly... this increased "efficiency of addressing"  is
actually an overall decrease of efficiency of the system.



There is mathematically no way to increase the total number of IP
addresses within the addressing of IPV4 protocol.   And the addresses
required for the network only increases.
Either the pace of IPv6 transition will rapidly accelerate,  OR the
pace of    introduction of new IPv4 networks will rapidly tend towards
zero.



--
-JH