Hello,

We have been experiencing high (~50%) CPU usage with Pingplotter Pro and it is making the GUI interface slow and unresponsive. When clicking on a different trace tab, it takes about 3 seconds for the new tab to appear.

The computer is a dual core 2.5ghz with 4 gb of ram.

We are pinging five hosts with each trace at a trace interval of 2.5 seconds, samples to include at 34560 (one day) and maximum samples held in memory at 103680 (three days). Pingplotter is running as a service and the web site is enabled.

I do notice that by stopping the trace does appear to speed the computer (and ping plotter) up again.

I also notice that tracing just the final hop doesnt appear to help, the CPU usage is still high.

Could our settings be causing this issue? If so which setting would you suspect is causing this? It seems like a computer with these specs should be able to ping far more hosts.

Thanks in advance.
-Matt

You shouldn't be setting this, but there are some condition where I can imagine it. What is your "Samples to include" set to? How many time graphs do you have showing?

It may help to take your current configuration and send it over to us (via the Help -> Email PingPlotter Support... menu) so we can have a look at it for possible performance problems.

We have customers running 100 targets+, so 4 shouldn't be a problem.

- Pete

Pete,
I sent the email from the help menu of pingplotter and received the confirmation.

One other thing that I would like to mention is that the computer running Pingplotter Pro is actually a guest virtual server on a VMWare ESXi server. The server is pretty beafy with dual quad core 2.5ghz xeon processors, 16gb of ram and 10k SCSI hard drives. So I cant see the issue being due to lack of resources.

Pete,
Any Ideas?

Heres a bit more info. When the server is configured for dual cores, ping plotter uses 100% of the second processor. When the server is configured for single processor, pingplotter uses 100% of the only processor.

Not sure if this helps.

False alarm problem found.

I did an upgrade on our VMWare server and went from version 3.5 to 4.0 and the problem went away.

Now the CPU utilization is from 0 to 20%.

Matt, thanks for the followup here.

We built an environment (VMWare Workstation 6.5) to test your scenario here, using your workspace, and didn't have the same performance problems you did (although we didn't have a full set of collected data either). After letting it run several days to get collected data, there was still no noticeable performance problems.

It's good to hear that the VMWare Server upgrade solved the problem.

We have noticed that VMWare sometimes "colors" the results - where the latencies will vary +/- some milliseconds (sometimes as much as 75 or so) based on CPU load. Because of this, running in a Virtual environment isn't really suggested.

- Pete