Re: PI addressing in IPv6 advances in ARIN

["Durand, Alain" <Alain_Durand@cable.comcast.com>]

Wrt Jordi's proposal:

I have sympathy to the idea of balancing PI need with routing table growth, however:

A) if PI addresses are to be returned at some point in time, they loose a dreat deal of their value. Folks like PI because it shields them from renumbering.

B) any address reclaim process might be lenghty and costly

C) given how long the shim6/multi homing has taken so far, it seems hazardous to make any bet that in 3 years it will be finish, implemented, adopted, deployed...

D) I am sensitive to the argument that v4 has not "melted" with PI, so why should v6 melt more? I believe the benefits of handing out PI *now* outweight the cost of a *small* swamp. 

E) given the state of v6 deployment now, I do not think there is much risk with this policy at this moment. And as Thomas relayed, this policy could be changed anyway if things get out of control.

F) if this means giving advantage to the early adopters by offering them PI, I look at this as a positive thing.

G) a key thing is to limit the size of the swamp that would be created, or more specifically limit its growth. So it might be a good idea to have a sunset clause in the policy, like it will have to be revisited in 2 years or when a certain allocation (or allocation rate) threshold will be reached.

    - Alain.