sysstat - Solaris' key system statistics at a glance
is thought to be a complementary tool in the list of
utilities included in Solaris for system observation (prstat, vmstat,
iostat, mpstat, nfsstat, netstat, ...). It provides an overall view of
the current system performance. Its key feature is to present all most
important performance metrics to the user at a single glance. Use it to
get a general idea of what is happening on the system. If you encounter
any bottlenecks, please ask the standard board utilities for more
detailed statistics to find out what is going on. sysstat
point you into the right direction, but it probably won't be able to
give you all the information you need in more complex scenarios.
Its output looks like this:
is stable and well tested, but if you encounter an issue
or miss a specific metric, I'd be glad to here from you.
is beeing developed on Solaris 8 and 10. It is targeted
for all releases starting from 8 (i.e. 8,9,10, and SX). I'd be
interested in reports concerning earlier releases of Solaris. If needed
I'll also patch sysstat for earlier releases, but I am and won't be able
to test other releases than Solaris 8 and 10.
sysstat is highly unportable because of the way statistics are
collected. I have no plan to support other operating systems.
For sysstat to work correctly in LDOMs, please install patch 137137-09.
Sysstat has a new, non-obvious feature: it has the capability to
monitor multiple machines on the network
. Just start a
sysstat daemon on every machine you want to monitor. Then start a
sysstat anywhere on the network in foreground mode, and all machines
will show up on a list at the bottom. You can now toggle through the
available hosts, using the hotkeys h
Comment on top on Solaris
Top is a popular tool on many UNIXes and UNIX like OSs. I cannot comment
on other OSs, but regarding Solaris its value is quite limited. First,
Solaris comes with prstat
which is much more versatile concerning
the analysis of running processes. Second, the summary statistics shown
by top are inaccurate concerning swap. Top gathers its swap information
using kstat's unix/system_pages, which really concerns total virtual
memory and not swap space. This has the negative effect that most people
think that Solaris uses a lot of swap space, when it is actually using
The right way to gather swap statistics, is to look at kstat's values
for unix/vminfo, which includes the correct numbers. Sysstat does this
correctly, and additionally, it shows a vast amount of statistics that
are unavailable on most other OSs. Please compare yourself.
If you are still not convinced, remember that top doesn't have a
networked mode. Sysstat in contrast supports multicasting, which allows
you to monitor multiple machines from a single sysstat session.
Please visit opencsw.org
for all kinds
of great binary packages of OSS for Solaris. A package of
Version 20111025 (current release)
Version 20091228 (withdrawn due to regression)
Version 20091024 (license updated to GPLv3)
Version 20060418 (initial release)
GNU GPL 2