Why do I not see any packet losses with "Samples to include" set to 1, but numerous packet losses on a sample setting of 2 (Trace interval = 5 seconds) ?

A second question: When I put an alias after the url address in the "Address to Trace" window, I get an error message when I start the trace. Am I missunderstanding this description of how to add names to the address: http://www.nessoft.com/kb/23 ?

Thanks,

Don

While you're collecting data, the "Samples to include" set to 1 is pretty volatile. As soon as we start a trace, we're looking at the *new* data for the current sample, which may not have arrived yet, so depending on your trace interval, there may be times when you never see that packet loss with samples to include set to 1. This is because we wait for 10 seconds for a sample to return, by default, and by the time 10 seconds is past, you might already be looking for the *next* sample (if your trace invterval is < 10 seconds). This can even happen with samples to include higher than 1 - if your trace interval and packet timeout period are such that a timeout doesn't arrive till after we've sent out other samples. With a trace interval of 5 seconds, we're never going to get any timeouts before the next trace happens.

There are several ways to solve this.
1) Decrease the timeout period (this is slightly different depending on which product version / edition you're using - if you need help, please ask).
2) Increase your "Samples to include" to 10 or so.
3) Increase your trace interval to > 10 seconds.

I would *highly* recommend increasing your "Samples to include" because 1 is really only useful when looking at history. I don't recommend relying on #3 because even if you set your trace interval to 11 seconds, then there will only be 1 second out of 11 where a timing out hop will show packet loss.

On your second question. Aliasing an IP only works with IP addresses - not names. So you can enter "192.168.1.5 My Server", but you can't enter "www.myserver.com My Server".

- Pete

Pete,

Thanks for the answers. I only tried setting the samples to "1" out of curiosity. I understand now why I wouldn't normally see the packet losses under those conditions.

Your answer to my second question is clear.

Thanks again for a great product. My problem is that I only turn to PingPlotter when I'm having problems, which is not too often. So first, I have to re-learn alot of the concepts and second, since I don't use it when "times are good", it's not quite clear to me what "good" traces are relative to the ones I see when the problems occur.

Also, I sometimes find it difficult to tell whether slow page load times are do to the packet losses I'm seeing, or issues at the server where the page is stored. If latencies to a site are 100ms and it takes 10 seconds for a page to load, I guess the problem is with the server?

Don