Here is an example of a PingPlotter graph. This data has some interesting characteristics that become relatively easy to spot when graphed.

This is a graph of 48 hours of data. One of the most interesting pieces here is the packet loss and latency changes over time. Both packet loss and latency look relatively good around 3am, but around 8am, things start to look worse.

Both packet loss and latency start to degrade, and in the evening hours we're getting nearly 25% packet loss and almost double the latency we do at 3am.

This time-dependent latency and packet loss changes are typical of bandwidth problems. During heavy load periods, performance degrades. The question, then, is whose problem is it?

If you have a look at the upper graph, hop 8 is showing no packet loss, while hop 9 is showing significant loss (the upper graph is showing averages for only 24 hours, although you can focus in on any period for any period of time). Hop 8 is on the Roadrunner network, while hop 9 is on Alter.net. This is a junction point, and the link between these two providers is probably not big enough. The subscriber should contact RoadRunner in this case to encourage them to solve this problem.