First off, you are absolutely correct in that that the min/max/avg are for the time period you specify. In your case here, you're displaying min/max/avg for the 10000 samples for about 5 hours. Now all of those 10000 samples need to be displayed on the lower graph - in about 400 pixels of screen space (rough estimate) which means that you have roughly 25 samples that need to be represented by every pixel. Ping Plotter averages all of the samples in that 1 pixel-width time period to decide what to display there. If you have one sample of 385 ms - and 24 samples at 7 ms, you're going to get something that looks like about 25 ms on the lower graph. As you change the scale of your graph, you'll see a lot fewer samples per pixel - and this maximum number will become closer to being represented 1:1. Note that packet loss works the same way - you can see some comment on that here:<br><br><A HREF="http://www.pingplotter.com/forums/showflat.pl?Board=Beta&Number=412" target="_new">Discussion about packet loss display</A><br><br>Now, as for your question about the spike on hop 2, but not hop 3 - yes, this is somewhat normal, especially under load. Based on this data, I can't say for certain whether the 1% packet loss you're seeing in hop 2 is being carried on to hop 3 either. This kind of thing is too normal - where an intermediate router is configured to down-prioritize ICMP echo requests (particularily, as you're seeing here, ICMP echo requests that reach it with a TTL=0 - as trace route and Ping Plotter use). In this case, all you can really do is ignore the badness you're seeing in hop 2. You do know that there is at least 5% packet loss that is happening between hop 2 and hop 3 - and it's all happening during load times of 6pm and 1am. This is very very standard for ISPs that don't have enough bandwidth - or possible problems in the connection.<br><br>The graph you're showing here is actually pretty effective - you could send this to your ISP and asking them why you're seeing packet loss percentages go up during peak periods. Even more compelling would be to collect several days worth of data and show how it consistently happens during peak periods.<br><br>If hop 1 is on the ISP side of your connection, then you have a very solid connection to them - and the problem is somewhere in their network (not a problem of multiple subscribers over-using the local shared cable bandwidth). If hop 1 is your cable modem - inside your home/business, then your connection is being affected by multiple subscribers using the shared cable bandwidth - but the packet loss is still primarily coming from something outside this connection - and something inside their network. Only they will be able to tell you for sure what the problem is.<br><br>In any case, you're on the right track. Of particular interest is the marked decrease in latency that happens at about 1:30 am or so - it's a very rapid decrease - more so than I normally see. The fact that this decrease is exactly reflected by hop 2 and hop 3 indicates that the problem is likely at hop 2, or between hop 1 and hop 2.<br><br>It's always challenging trying to prove that there's a problem. Start sending these to your ISP on a pretty regular basis. I'd suggest saving your images as .png files instead of .bmps (they load about 10 times faster), but it looks like you're on the right track with your web site and images there.... (Those 48 hour graphs are just sweet for showing the times when there's a problem!)<br><br>Keep us informed as to your progress here! Good luck with your ISP!<br><br><br>
Previous Topic
Index