Re: [homenet] Let's make in-home ULA presence a MUST !?

[James Woodyatt <[e-mail]>]

On Tue, Oct 14, 2014 at 2:49 PM, Ted Lemon <[e-mail]> wrote:

>
> I don't think the objective is for the ULA prefix to be invariant.   It's
> for the availability of a ULA prefix to be dependable, and for flash
> renumbering to be avoided whenever possible.   So there's no problem with
> deprecating a ULA when you have two, and no need for the ULA to remain
> stable over long periods of time.
>

But there is a problem with only deprecating prefixes without expiring
them. If they never expire, then they accumulate without limit within
existing networks as they join with newly commissioned networks over the
course of their lifetimes.

The reason to want there to always be a ULA is that if you use a GUA as a
> ULA, the life cycle of your home network numbering is out of your control,
> and in the hands of whoever gave you the GUA. That's the only thing I think
> the ULA prefix has to do on a homenet: provide you with dependable,
> graceful homenet-local numbering.
>

So what's the problem? My language above ensures that home network hosts
always have at least one gracefully renumbered IPv6 address routable
throughout the entire network. If we need a further guarantee that hosts
always have an *invariant* address— which is an objective you've said above
that you think we don't actually have— only then are we faced with the
problem of prefix accumulation through network joins, which is a problem
I'm not sure we know how to solve effectively. My proposal avoids that
trouble.

--

To answer a previous question: I would say the reason I thought it worth
expressly zeroing the preferred lifetime on the locally generate ULA prefix
when another prefix is advertised is to expedite the transition to
preferring any delegated ULA prefix over the locally generated one.
Admittedly this is perhaps not worth the effort, and I won't argue further
for it.

-- 
james woodyatt <[e-mail]>
Nest Labs, Communications Engineering