On 4/24/2015 9:33 AM, Templin, Fred L wrote:
> Then, instead of saying "shut the tunnel down" you should say "shut the path
> down". 

A tunnel is defined by its ability to transfer packets between ingress
and egress. How the tunnel decides what loss rate is sufficient is its
decision.

A tunnel has no way to know "path" at all; that concept doesn't exist in
the Internet except using LSR or SSR. What it does know is loss rate and
whether that correlates to packet properties - flow IDs, ports (if
visible), packet length, etc.

Any of those can be a consideration in deciding when to declare a tunnel
down. Short of sending every bit combination repeatedly and using
positive acknowledgement, an ingress can't know whether the packet
contents are causing loss.

However, since I don't hear anyone else on this list raising this red
herring, it doesn't seem critical to even explain in detail as to why it
need not be addressed in this doc.

Joe



_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to