Pete,<br>While the packet loss graphs look good, it is not helpful to sidetrack onto pixel widths when you are trying to convince someone that there is a problem in the network.<br>Couldn't you graph (behind the black time plot) the rolling percentage of non-returned packets in the current "samples to include" period, possibly onto a fixed log scale of minimum .1,1,10, maximum 100%.<br>That would be easy to follow even for some of the people I have to talk to... then again I'm always over optimistic.<br>Many thanks,<br>Ed<br><br>