I am one of the n users of a shared unix machine. For reasons unknown, the machine is not "responsive" enough. For example, it is slow on interactive commands, it takes few noticeable moments for any action (e.g. mouse movement, editor (e.g. gvim) keystrokes) to be visible. The problem is, the people supposedly responsible for addressing the issue do not agree that the machine is not responsive. They do some few simple things and say, "It works fine!"

How can responsiveness be quantified? What (all) can I measure?

I can run shell commands (e.g. top) periodically with cron and collect statistics, but I am clueless regarding what is a good statistic to go after.


I connect to the machine over VMC.

4 Answers 4


Are you connected directly to the machine or are you logged in remotely via ssh or the like? If the latter, make sure that the perceived lag isn't caused by the network...

If connected directly, besides looking at IO and system load, you might have a look at memory usage. If a system starts swapping heavily, the perceived responsiveness dramatically decreases. The command "free" shows how much memory is available and how much of it is used.


This isn't strictly speaking the same as "responsiveness", but one metric you should probably check is the system load average; uptime will show the average over the last 1/5/15 minutes:

$ uptime
02:30:33 up 6 days, 6:30, 12 users, load average: 0.85, 0.65, 0.57

A high enough load will perceptibly slow the system down

  • Accepting this since uptime suggestion is simple and useful.
    – Arun Saha
    Oct 24, 2010 at 22:23

If many people work on a system IO can become a big problem: When starting a program the system might have to load stuff from the harddisk (program files, libraries, resources) and the memory might be full as well (leading to swap being used heavily). Harddisks are slow. As in really slow.

There is a program called "iotop" that allows you to track which user uses how much IO and how much disk IO you have overall. Might help you get some material to prove to the admins that the system is in fact running badly and why.


Servers are usually built for high throughput instead of high responsiveness, so you should keep that in mind when evaluating the responsiveness of a server system.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.