Atop was absolutely my favorite tool when I was supporting lots of interactive systems and users complained to me that "foo is slow". Even better, the logging l

Atop was absolutely my favorite tool when I was supporting lots of interactive systems and users complained to me that "foo is slow". Even better, the logging let me determine the cause after the fact when a user complained "foo was slow". The beauty of it is that it's using process accounting to gather the resource utilization of everything that was running on the system between two points in time. Compare that to a more standard metrics-gathering system like collectd, where I would have to know the names of the processes I want to monitor in advance (http://collectd.org/documentation/manpages/collectd.conf.5.shtml#plugin_processes http://collectd.org/documentation/manpages/collectd.conf.5.s...). The case study at http://www.atoptool.nl/download/case_leakage.pdf http://www.atoptool.nl/download/case_leakage.pdf is worth a read to get a feel for what it's like to use atop.