[c-nsp] something a little different rfc1918 from transitnetworks?

[Danny McPherson]

On Nov 13, 2007, at 9:04 AM, Gert Doering wrote:

> Hi,
>
> On Tue, Nov 13, 2007 at 10:29:28AM -0500, Eric Van Tol wrote:
>> I think the question the OP is asking is, why would the same exact
>> address be seen from two different peers, each 3 seconds apart?
>
> Equal-cost paths, load-balancing SYN and retransmit over two different
> transit providers...

OR perhaps even a traceroute (or PMTU, or other) client downstream
eliciting time-exceeded, fragmentation needed, or other messages from
non-adjacent upstream networks that are using RFC 1918 space on
"internal" links.

Given the ACL is denying them, just another reason why employing
RFC 1918 in this manner is a bad idea, and PMTU today is mostly
broken.

-danny