I'm working on a project that one of its tasks is to measure the average memory and cpu usage for the passed X minutes. The plan is to cover tier 1 Unix/Linux OS,
I'm a programmer and not very familiar with the different technical aspects of cpu and memory but I do know that these is no single number that represents them.
I found out in Unix the command "vmstat" that gives information about memory, cpu, page fault and other system parameters but I'm not sure this is the right tool. I mean, there is user cpu usage and system cpu usage. How do you figure out a single number that represents the overall cpu usage? The same goes for memory, there is phisical memory, and virtual memory and some other memory related parameters.
And what happen if you have 16 cpus? Can you boil it down to a single number?
The design is that a user will define a minimum free cpu usage and a minimum available memory, and if these conditions are met, only then other tasks will be performed. This is to prevent a situation in which a task is submitted to a machine that is already overloaded.
How would you approach such a task?
Do you know of any 3rd parties softwares or utilities that yield such information?