Rodrigo:
There really is no such thing as a broad-ranging "normal" for *most*
system statistics, except as one might observe specific empirical norms
over time for a specific workload running under a specific
hardware/software configuration.
Presumably, you are looking at these numbers for some clue regarding why
some application is not performing well. My advice is that it is far
easier to first determine whether or not the hardware and software are
well-configured for your application with respect to known Best
Practices - and far more practical to do *that* analysis before delving
into low-level OS statistics. Indeed, most low-level statistics are
rather meaningless absent that context. Thus, it is critical to know
exactly what you are trying to accomplish, and by what business metrics
you are measuring your success. This includes knowing the hardware, the
OS version and patch level, the application, and their major
configuration parameters.
Assuming the system and application are well-configured, 'mpstat' might
gives some hints about where to investigate for performance issues, but
other tools would be indicated for drill-down analysis. 'prstat -m' and
even 'ps -o' are among the most useful and basic ways to begin gaining
insight into workload dynamics, but the tool set in Solaris goes very
broad and very deep. That's why the kind of broad question you ask
frequently draws a return volley of: "What problem are you trying to solve?"
Best regards,
-- Bob Sneed
Rodrigo Nascimento wrote:
Hi All,
I'm try to understand the output of mpstat command, but whether I really want
to understand it I need to know what are considered normal values.
CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl
0 22 395 26554 3344 2855 2958 700 773 270 0 4762 41 36 23 0
How can I say that 773 for migrations is a good value for this metric if I
don't know what is considered normal value.
Maybe anothers factors will need to be consider, as What kind of program runs
in this machine. Is cpu-bound or I/O-bound same thing like this, but by the way
I believe that exists a standard.
Thanks and regards,
Rodrigo
This message posted from opensolaris.org
_______________________________________________
perf-discuss mailing list
perf-discuss@opensolaris.org
_______________________________________________
perf-discuss mailing list
perf-discuss@opensolaris.org
