One thought: There seems to be an indication (in the description of
"Path-Set TROs") that results for an interval are only used if the Master
was able to measure both directions of all A->X..->Y->B paths involved.
If i'm interpreting this correctly, then this seems like it will throw
away some useful info. Even if the Master doesn't get *all* the data it
would like to, it's still useful to know that there were *some* better
paths during that window.
neal
On Mon, 17 Aug 1998, Stefan Savage wrote:
> Also, it'd be nice if people could take a moment to read John's second
> measurement proposal (its only 3 pages long) and comment on it this
> meeting. This is important stuff... these numbers are the hottest thing
> we've got, and the strength of our methodology is going to directly
> affect how much people believe us. Anyway, John's proposal is at:
> http://www.cs.washington.edu/homes/geigudr/tiber/Strobe.ps