Re: [v6ops] (re)numbering [ULA draft revision #2 Regarding isolated networks]

[Mark Andrews <[e-mail]>]

In message <CAEmG1=rz=[e-mail]>
, Matthew Petach writes:
> 
> On Wed, May 28, 2014 at 2:24 PM, Mark ZZZ Smith <[e-mail]>
> wrote:
> [...]
> 
> > RFC1918s have provided that internal connectivity robustness to both home
> > networks and enterprise networks. Of course the drawback is that in IPv4 it
> > is binary - hosts either have RFC1918s or public addresses, so if you have
> > RFC1918s you have to use NAT to access external destinations on the
> > Internet.
> 
> Wow...that's news to me.
> 
> For a decade now, I've been using
> RFC1918 addresses+global addresses
> in IPv4 on my home network; each
> host has an address from each subnet,
> and uses the 1918 addresses to reach
> internal-only devices (printers, terminal
> servers, etc.) which only have RFC1918
> addresses, and use the globally routed
> IPs for reaching non-local destinations.
> 
> I'm not sure I'd agree with your characterization
> that IPv4 is different from IPv6 in that regards;
> there's nothing in the IPv4 world that prevents
> hosts from having multiple addresses, and
> making use of them.
> 
> It's definitely a plus to have internal connectivity
> stay working regardless of external connectivity,
> I completely agree with you on that.
> 
> Matt

It may work with some machine some of the time.  It is not guarenteed
to work with all machines all of the time.  I've definitely used
machines which didn't support multiple IPv4 addresses on the same
interface.


-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]