> > "Why measure packet loss percentage?" If it is not in an attempt to
> > predict future behavior, then it seems to lose much of it's value for
> > anything beyond repairing broken infrastructure - it doesn't seem to
> > tell you anything of much value for use in capacity planning, for
> > example.
> The way from current packet drop to current total bandwidth need on
> just a single path would appear to be non-trivial. The way from
> current packet drop to future capacity need would seem rather more
> complex.
> Not that these are unworthy goals. But perhaps there are some closer
> goals which might be more easily achieved first such as, as you say,
> correlated QoS measures.
Randy,
I agree that going from current drop to a new value for how much bandwidth is needed is extremely non-trivial. But it seems that going from current drop to the realization that more bandwidth is needed. :-)
-------------------------------------------------- Jon Boone Operations Engineer ISC Networking University of Pennsylvania tex@isc.upenn.edu (215) 898-2477