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

[Michael Richardson <[e-mail]>]

Lorenzo Colitti <[e-mail]> wrote:
    >> I think that we should design our protocols to work independent of
    >> what prognostications we are able to make about what users might do
    >> today, rather than making decisions that will ensure that certain
    >> perfectly valid uses of the network will fail tomorrow.

    > But you're saying you want ULAs because you want to continue to do what
    > you were doing yesterday: persistent connections, like SSH and
    > X-windows. I think you're trying to fix the problem at the wrong
    > layer. But I don't expect we'll agree with me on that.

It's not about whether or not SSH and X-windows can reconnect or not.

It's that, even if they try, the target host has been given a new address,
and there are long-lived credentials which have been tied to the name
and the address.  We did this in the days when we didn't have mDNS, and
we couldn't trust that well DNS, and putting the IPv6 address (the ULA),
of that ssh-enabled powered bar into my /etc/hosts (or .ssh/config or
registry, or shell script) seems like an entirely sensible thing to do.

I will go back and read James' message about joins and splits.
It seems that we have this problem with GUAs as well, and it seems that
the whole address selection issue exists without ULAs, as long as one has
multiple ISPs.

And LLNs will be installed with ULAs, and in some cases, we might really
want to route that into the homenet, so ULAs show up even if CPE devices
don't create them.

{BTW: I've never been a fan of ULA-Random, and I really really would like
to fine some way to make non-routable IPv6 globally unique addresses, that
include reverse DNS and whois, available for a low enough price that nobody
has to think twice about getting them.  But, in this case for the home of the
unsophisticated user, ULA-Random, seems correct}


--
Michael Richardson <[e-mail]>, Sandelman Software Works
 -= IPv6 IoT consulting =-