[6bone] non-global address space for IXs (was: 2001:478:: as /48)

Pekka Savola pekkas@netcore.fi
Sun, 7 Sep 2003 08:08:06 +0300 (EEST)


On Sat, 6 Sep 2003, John Fraizer wrote:
> On Sun, 7 Sep 2003, Gert Doering wrote:
> 
> > Hi,
> > 
> > On Sat, Sep 06, 2003 at 06:37:27PM -0400, John Fraizer wrote:
> > > On the other hand, if that
> > > address space is not globally routed, it breaks PMTU-Disc, traceroute,
> > > etc.
> > 
> > It does nothing of this, *unless* you're also doing reverse-path filtering
> > on your external links (which is a dangerous thing in most cases anyway).
> > 
> > It breaks pinging / tracerouting *to* a specific router on its IXP
> > address, indeed, but not *through* the router, which is by far the
> > most common usage.
> 
> 
> Sorry Gert.  You don't router IX space and you wind up with the same
> issues as are presented by the boneheads who use RFC1918 address space on
> VISABLE links in V4 space.  *EVERY* link in the chain has to be able to
> participate for PMTU-D to properly function.

Every link where PMTU changes, you mean.  Every link which doesn't use 
e.g. loopback addresses to send out these ICMP messages, you mean.

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings