[arin-ppml] ARIN's Authority - One view (was: Re: LRSA concerns)
tvest at pch.net
Mon Aug 25 20:14:03 EDT 2008
On Aug 25, 2008, at 6:27 PM, William Herrin wrote:
> On Sat, Aug 23, 2008 at 1:29 PM, Howard, W. Lee
> <Lee.Howard at stanleyassociates.com> wrote:
>> What can we do to ease the transition to IPv6?
> Not much. There are three key obstacles to IPv6 deployment:
> 1. In a down economy, deploying and maintaining both IPv6 and IPv4
> costs noticeably more money than just IPv4 and it is not practical (at
> an individual or org-by-org level) to discontinue the IPv4 costs after
> deploying IPv6. Worse, it has the costliest of costs: manpower from
> your top technical people diverted from other tasks. Hence any
> organization which deploys IPv6 is economically disadvantaged versus
> those competitors who don't.
Wouldn't competitive pressures be just as relentless and unforgiving
in an up economy?
I'm trying to think back to what ambitious undertakings were achieved
back in the go-go days (c. 1998-2000), but I can't remember any.
If I'm not too off-base here, it's not an indictment of operators, or
competition, or capitalism, or anything -- just an observation that
some kinds of problems cannot be solved through purely uncoordinated/
competitive action alone.
If deploying IPv6 is one of those problems, then it seems to me that
we'll probably be facing this or an equivalent problem sooner or later
-- at which point if we've lost the coordination mechanisms that we
have today, they'll just have to be reinvented.
> 2. Some damn fools at the IETF recommended that applications which
> support IPv6 should use it in preference to IPv4 when available.
> Nearly all of the application developers have followed that
> bone-headed recommendation. So as soon as you turn IPv6 on in your
> network, -everything- in your network tries to use it first and only
> falls back to IPv4 if IPv6 doesn't work. Since that detection and
> fallback process is considerably less than perfect, the net result is
> an instant drop in overall system reliability when you deploy IPv6
> (since the v6 parts of both your software and the network overall are
> less well tested), making IPv6 deployment undesirable.
> 3. Based in part on the recommendations from the same knuckleheads in
> #2, the registries (including ARIN) determined that a mid-90's style
> Internet land rush would undesirably generate the same
> legacy-registrant problems in IPv6 that they have with IPv4. So,
> policy was set such that we would not have a land rush. Hence no rush
> to deploy IPv6.
> Of these, only #3 is addressable within ARIN's bailiwick. Since last
> year, ARIN has at least removed the incentives for folks to work
> against IPv6 deployment (ALL present registrants now qualify for a
> direct IPv6 assignment or allocation). However, without a land rush
> folks will tend to deploy needs-justified resources only as they need
> them. For better or for worse, nobody needs IPv6; they need more IPv4.
> IPv6 functions as a substitute about as effectively as tap water at a
Well, in that case everybody had better get ready for last call, if
not next year then soon enough.
We don't have to go home, but we can't stay here ;-)
With promises of fewer postings tomorrow,
More information about the ARIN-PPML