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

Re: Issue 3: on link assumption considered harmful



jrh@it.uc3m.es said:
> For example, your fridge wants to talk with the scheduler to ask for
> more food. The oven, which is sending RAs, is turned off.  This will
> cause no communication at all. I prefer 3 seconds of delay. I'm the
> auto-communication kind of man.

I don't see how you reached the conclusion that the removal of the
on-link assumption would cause lack of communication in your example.
If you're depending on a fall-back to IPv4 (and I assume you are since
you mentioned a 3 second delay), then communication will occur more
quickly without the on-link assumption...  Can you explain your
reasoning?

Your example is not well defined.  Are the fridge and the scheduler on
the same link or not?  Do they have manually configured IPv6 addresses
or statelessly autoconfigured addresses?  Do they have IPv4 addresses?
Is the oven both the IPv6 and IPv4 router?

> 
> I don't intend to make you laugh but I think I will have to get some
> information about the zero-conf WG's jobs before going on  this
> discussion, because I thought that this feature was a GOOD THING.

The zeroconf requirements don't include the on-link assumption.  They
also don't include nodes residing on the same link, but it's not clear
if the example you gave depends on that or not.

> > The more obvious problem you may not be seeing is when an implementation
> > has enabled dual-stack but does not have IPv6 connectivity at all (i.e.,
> > only the link-local addresses). 
> 
> This is the common case nowadays. When I connect to Internet at home with
> my dial up provider, I've got IPv4 and also a link-local address and I've 
> never experienced any problem at all.

Your experience is anecdotal.  There may be two reasons for your not
experiencing problems.

1. In general, do you attempt to communicate with nodes whose names
   resolve to both IPv4 and IPv6 addresses?  Maybe not.

2. Some OS's never implemented the on-link assumption to begin with,
   maybe yours is one of them.

-Seb


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