An example is the cdm probe that gathers cpu, disk and memory by using system/OS level calls. Depending on the flavor of UNIX, the method of gathering the metrics may provide a different answer than a different method.
In order to evaluate the value of the probe-metrics, an understanding of what OS level call is used, the translation, time period and reporting cycles would need to be understood.
The details for how a probe gathers the metrics should be part of the documentation set for that probe, since without it you are trying to run the probe and have multiple putty windows open running the various shell commands in a loop to try to determine which of the various commands are giving you the right unit, scale, time frame and context.
I had to create a support case to request this information since after pouring through the documentation was unable to locate the information.