[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Appeal on "Unique Local IPv6 Unicast Addresses"



Alain Durand <Alain.Durand@Sun.COM> writes:

> On Feb 27, 2004, at 5:23 PM, Thomas Narten wrote:
> >
> > Let me ask you this then. If the word "permanent" is not appropriate,
> > what word is? To me, "not permanent" means that at some future time an
> > allocation that has been made to an endsite may be revoked.

> The point I'm arguing is that it is not the IETF business to decide 
> this.

I respectfully disagree. Seems to me that whether or not assignment
should be permanent is a property of the type of address we are
defining and goes to heart of why we are defining unique local
addresses. My understanding is that end sites are supposed to be able
to use them however they see fit, and specifically will _not_ have to
worry about someone reclaiming them or taking them away. If an end
site has to worry that it may have to give up the address at some
future time, these addresses have just become a lot less attractive
(or useful).  Thus, it is important that the assignments effectively
be permanent.

> So not saying anything is probably the best option. Another one is
> saying that the entity in charge of the allocation is required to
> provide those allocation for the long term and not specify anything
> further. It is a policy decision made by the entity(ies) in charge
> to decide what long term means.

Letting the entity that operates this service decide the terms of the
address (how long an assignment is good for, under what terms it
should be reclaimned, etc.)  without any sort of guidance is a recipe
for creating a bad situation that we cannot fix. I.e., we get a
service that isn't what this WG intended and that we can't fix it
because we don't control it (e.g., in a worst case scenario, a newly
created monopoly accountable to nobody).

> See the contract between the chosen entity and the customer.
> Again, this is policy, not protocol. i.e. this is not IETF business.

Are you really suggesting that some third party (not under IETF
direction) define the policy, even if the result of the policy could
defeat the purpose of having such addresses?

> >  Is
> > there some future scenario we're worried about where it becomes
> > important to be able to reclaim these addresses? I.e., are they ever
> > going to become a scarce resource?

> I'm worried about the scenario where IETF does policy and not protocols 
> anymore.

I believe you are confusing properties with policy. One can't
completely separate the two. If you look at CIDR, we've moved to
address leasing because we don't know (technically) how to do global
routing if all address assignments are permanent. In the case of
unique local addresses, there are no technical issues with making such
assignments permanent (or do you disagree?). Thus, there is no reason
why we can't make such assignments permanent. (Note: I'd agree with
you that the assignments shouldn't be permanent if there was a case to
be made that it may become necessary to reclaim them at some future
time. Is there?)

Thomas

--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------