[arin-ppml] Ending point to point links as a justification for a /30?

[William Herrin]

On Thu, Jul 29, 2010 at 4:25 PM, Richard A Steenbergen <ras at e-gerbil.net> wrote:
> Rest assured I know exactly how traceroute works. :)
>
> Sourcing the ICMP messages from a real loopback IP solves the problem of
> unreachables not making it back to the host due to filtering, but it
> still completely disrupts traceroute as a diagnostic tool. There is a
> lot more to troubleshooting a network than knowing the packet went from
> router A to router B, which is all you'd be left with if you sourced
> your icmps from a loopback.
>
> For more information on how network operators use traceroute to
> troubleshoot real issues, take a look at:
> http://www.nanog.org/meetings/nanog47/presentations/Sunday/RAS_Traceroute_N47_Sun.pdf

Hi Richard,

I notice that your "factoid" on slide 10 that obeying RFC 1812 changed
from "would prevent traceroute from working properly" in your earlier
presentation to "would completely change traceroute results." The
latter observation is correct though the "completely" still overstates
the case.

The use of any number of technologies changes what traceroute tells
you. I'm sure you've seen the MPLS networks that show only two hops
revealing no internal information. The more and more frequently
employed tunnels and VPNs affect traceroute's function. And traceroute
never could tell you that the set of switches between two routers was
acting up.

This too would change what traceroute tells you. But at least it would
tell you something, which is often more than you get now when a router
is configured with RFC 1918 addresses.

Like your presentation by the way. Very thorough. Especially the
points about latency and asymmetric paths; so many folks don't get
that.

Regards,
Bill Herrin

-- 
William D. Herrin ................ herrin at dirtside.com  bill at herrin.us
3005 Crane Dr. ...................... Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004