> Date: Wed, 24 Jun 1998 20:16:06 -0700 (PDT)
> From: Eric Hoffman <hoffman@cs.washington.edu>
> To: syn@cs.washington.edu
> Subject: re: probe loss rates, calculation of
>
>
> 1. The current TTL is not the max; then traceroute recieved a
> response from someone identifying themselves as the recipient. Take
> all failed probes at this TTL, plus all failed probes from all TTLs
> previous, where such a previous TTL had nothing resolve. (ie, * * *)
> 3. The current TTL is max, and the precise target resolved. Then
> treat this as case 1
>
> right. its important though that this is a fairly radical
> underestimate of the two-way loss since most of the packets didn't
> actually have to make it all the way to the destination and back
I disagree with this last statement; Given the following resolution,
* Q *
T * *
Where T is the target, the rate counted by the above statement is 2/3.
What I meant was that this:
* Q *
* * *
T * *
Generates a loss probability of 5/6. Why? Because we don't know if that
3-star line was due to an incommunicado router (unlikely, that close to an
endpoint), or just due to all the packets being dropped.
Does this restatement bring us into agreement, or am I missing your point?
>
> 4. The current TTL is max, and something resolved, but it
> wasn't the precise target. Then we don't know, as this could either
> be a logical proxy for the target, or it could be something else.
> Punt.
>
> I don't see any reason not to treat this as the same as reaching the
> final destination...but if this is a small number of samples then its
> probably best just to throw them out
My thought on this is, that the joint probability of us reaching maxTTL
and having the target T resolve to some nearby IP T' right at maxTTL is
much less than the probability that T is greater than maxTTL hops away
from sourceIP, and we simply got an intermediary.
>
> 2. The current TTL is max, and nothing resolved. Then take all such
> complete failures, moving back, until at least one IP resolved for a
> given TTL, this last non-inclusive.
>
> this case is fairly hard...since the loss rate at this instant was
> essentially infinite..if you express each of the traceroutes as a
> probability then 1.0 works...maybe it would be better if you tabulated
> ocurrences of total reachabilty failure seperately
I like that calculation. You're right; this is fairly infrequent. I'll
see about a separate tabulation, though.
Thanks,
John
_____________________________________________________________________________
"The human mind is a 400,000-year-old legacy application...and you expected
to find structured programming?" -- Randall Davis, 1996 AAAI Pres. Address