Re: [v6ops] PI heresy

[Owen DeLong <[e-mail]>]

> On Nov 12, 2015, at 16:15 , Mark Andrews <[e-mail]> wrote:
> 
> 
> In message <[e-mail] <mailto:[e-mail]>>
> , George Michaelson writes:
>> 
>> this is a specious argument.  Its as specious as the 'one IP per citizen'
>> complaints relating to some national numbering plans, or the model of
>> assigning v6 to motor vehicle vendors.
>> 
>> the reason its specious, is twofold
>> 
>> 1) it has a natural ramp up. to get every SME up to speed and *able* to
>> multihome and route, takes time. Therefore there is no cliff moment: its a
>> process. A process which emerges, over time, and permits adjustment of the
>> routing model to cope.
> 
> Your assuming multi-homing.  If you don't have PA, ULA + PA and
> don't have ULA + NAT then PI is all that is left and there is a
> very small bar to jump to be singly attached.  You don't even need
> to run a routing protocol.  You just need the ISP to inject the /48
> for you.
> 
> Owen wants us to just recommend PI and to forget about PA, ULA +
> PA and ULA + NAT.  If that is the recommendation then we will have
> 125M /48's within a couple of years.  That's why we should reject
> Owen's recommedation and go with something more conservative.

No… You are placing words into my mouth that simply were never there.

I want to forget about ULA and ULA+NAT.

I’m fine with PA for those willing to accept it and I advocate PI for those not
willing to accept PA and its limitations.

I’ve never campaigned against PA. I have campaigned against the refusal
of PI for those that did not want PA, but that is a different issue. PA where it
is an adequate solution is fine with me. IMHO, those are a fairly limited set
of circumstances and we should be expecting a relatively significant growth
in the IPv6 routing table over the next couple of years, whether we like the idea
or not. But again, that’s a separate issue.


> 
> 	e.g. a mixture of PI, PA and PA + ULA.

I accept a mixture of PI and PA. I even accept PA+ULA, though I argue it is an
inferior solution with significant drawbacks.

Nonetheless, anything involving NAT whether or not it involves ULA on a system
which is connected to the internet is a harmful “solution" where the harm inflicted
often (usually) extends beyond the network that chose to implement said “solution”.

Owen