[ppml] PIv6 for legacy holders (/w RSA + efficient use)

[Randy Bush]

Paul Vixie wrote:
>>> so, no other possible outcomes were studied, you "just like this
>>> one"?
>> did he say that?  what i heard him say was that simple business and
>>  economics will drive toward this.  made sense to me.
> i'd like to know what makes this alternative the most likely.

you may want to read drc's mail

> if someone who holds the "markets are efficient and inevitable so
> just relax" view can show how that view is falsifiable, i'm
> listening. 

i am not sure anyone but you has been talking about market efficiency.
market reality, maybe.  and nothing is inevitable.  well, death and
taxes, i guess.

but, unless we change the perceived costs to the enterprises, services,
and large last mile isps, what drc, i, and others have been describing
is indeed the most likely outcome, whether we like it or not.  the
underlying problem is that the decisions will be made by people counting
beans, not rosaries.  ever since the dot bomb, it's been all about this
quarter's net, not what's right for the internet (as if it ever really was).

right now, in most cases v4 and v4 nat look a lot cheaper to new
deployments and expansions than v6 with v6/v4 nat (alain has a good
counterexample, but it involves really massive scale).  this is because
v4 kit is a known reliable quantity, cheap, and with no compatibility
issues with the rest of today's internet.  it's a no-brainer.

the fun question is how expensive will the street price for a /24 or /19
or whatever of ipv4 space have to be to make folk perceive that the next
quarter costs of using ipv6 internally are less?  and note that the
perceived ipv6 cost has a high risk add-on, as it is not clear that all
the pieces of the puzzle are there for non-trivial cases (we should do
something about this).

randy