Ok, first you will need to ensure that you are collecting the necessary metrics. Then you will need to look at "Monitor Server Settings" to determine the level of detail you wish to retain and how long. You will see that the detailed data begins to "smooth" after a certain amount of time. Also, in your Monitor Agent Configuration, you will see that it will record metric values every x number of seconds. I have mine set to every 10 minutes, which is a bit aggressive, but I need to record my numeric values that often and upload them to the NS for historical performance detail/reference. I then use Historical Performance Monitor and encourage others in my org to do so to look at performance on a specific server. The graphical nature of this approach seems to provide the clearest and quickest insight into a server's performance. Also, as you highlight and zoom in on an area, (if you are collecting process values too), you can see what processes were running at that point in time.