> > Jim Bound wrote:
> > here is not reason at all why what you stated could not
> > be link-local addresses. I would argue if sensors do what
> > I hear they will do link-locals are fine and we do have
> > controls on that and they are not forwarded off the link.
>
> That's not the way it works. In the rather classic triple or quintuple
> system redundancy, each of the devices that can control something is on
> a separate bus (a separate subnet). But it might talk to the other two
> or four all the time, and they occasionally vote and might decide that
> one of the devices is out of whack, things like this. So, there are
> multiple links in case of cable failure or jabbering NIC or something,
> but they might talk to each other. Site-local.
doesn't follow. you haven't made a case for site-local, you've just
included the word at the end of the paragraph.
> But the plane itself would not be completely isolated either. Let's face
> it: directly or indirectly there is almost no network today that is not
> connected to the Internet not even a plane in flight. So saying that SLs
> can not be used in networks that are connected to the Internet is the
> equivalent of killing them.
killing SLs would not be a crime ... IMHO it would be a good thing,
though I don't expect that to happen. ("it was a mercy killing, honest!")
and the proposal I made was saying that they could not be used in networks
that were connected to the Internet using IP. connection via ALGs would be
acceptable. the point is that you don't expose a mixture of SLs and globals
except to special-purpose apps like ALGs. (now if we can only figure out a
way to make that stick this time)
> Same as the other examples I used before: sensors/control devices in a
> metropolitan water distribution system, or in a power grid. SLs are a
> perfect choice for these,
no they're not. because often the reason you want these sensors on IP in
the first place is so that you don't have to provide your own infrastructure
to talk to them at a distance.
> These topics have been discussed years ago, and I question why we need
> to revisit this.
because in all this time nobody has figured out a sane way to actually
use the things, even attempting to do so increases complexity without
a significant gain, and they don't solve the problems they were supposed to
solve. furthermore, appeals to past ignorance aren't exactly convincing.
Keith
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
