And so it ends...

Thu Feb 3 22:09:42 UTC 2011

On Thu, Feb 3, 2011 at 1:52 PM,  <Valdis.Kletnieks at vt.edu> wrote:
> On Thu, 03 Feb 2011 13:39:25 PST, George Herbert said:
>
>> It's probably most practical for them to renumber into a subset of
>> their existing space, collapsing down from the whole /8 into a /10 or
>> something longer, which would free up 75% of that space or more.
>
> And they want to go to the trouble of doing that, why, exactly?
>
> Imagine taking that to the CIO and/or budgeting people: "We want to start this
> $mumble-million project to renumber".  What's the first question they'll ask?
> "What's it mean for *our* bottom line?" What's the second? "Then why do we want
> to spend this money?"
>
> It just ain't gonna happen till  you have good answers to those. "We can spend
> $mumble-million renumbering into 1/4 of the space, and then sell off the other
> 3/4 to various entities for an estimated $mumble-million+20%".
>
> *Then* it will happen.

Some of them won't have to renumber at all to collapse into a subset
(from what I was told).  Some are spaghetti messes.

Putting out a policy best practice that says "You really should do
this, please" doesn't force multi-million-dollar projects, no.  But
might prompt returns where no renumbering is required.  And can
hopefully encourage network revamps going forwards to recover space as
they go, if it's not too painful.

The alternate method - to just openly commoditize it - will also work,
but will incur significant political pushback within the community.

I don't know which path is ultimately more productive over long
timescales.  I think that a best practice asking for handbacks is at
least harmless in the nearterm.  If we need time to overcome
opposition to commoditization on our side of the fence, then that
should start now, but we can't plan for overcoming that on a
particular schedule.  Given that APNIC hits their wall in 6-7
months-ish, I don't know that we can move quickly enough, but someone
needs to start and see what happens.

-- 
-george william herbert
george.herbert at gmail.com