System monitoring with nmon

Today I was working on an AIX server and I ran across a utility called nmon. It seems like top on steroids. Like top, it can report per process CPU and memory utilization, but also quite a bit more. It also seems to incorporate disk I/O, virtual memory usage, network throughput, NFS stats and more. And all from the terminal.

Since it seemed so potentially useful, I wondered if it was strictly an IBM tool. However it seems like there’s an open source version available on sourceforge:

It’s also available in the Debian repositories. RHEL/CentOS is a little different; it needs to be installed from EPEL.

Has anyone else used this? If not, what do you use for system monitoring?

This is from a relatively idle workstation. The layout of the display is not fixed; you can add and remove different graphs and stats as needed.


What does steal % mean?

Here’s an explanation:


Yep. Always have nmon installed. First encountered it under AIX, and have installed it everywhere ever since.

One other neat feature AIX has - a gooey for system management that tells you what the command-line to do the same thing is (SMIT). Very handy for working out actions ahead of the ‘change window’.

1 Like

Yeah, everyone at work types ‘smitty’. Just for fun I typed ‘smit’ instead, and was pleased with the results. Not a lot of window dressing, but very useful.

1 Like

@Goalkeeper This explanation is both less verbose and clearer:
Essentially it’s the amount of time your VM’s vcpu spends waiting to be serviced by it’s hypervisor.

1 Like