performance with sar

Carme Torca — Wed, 06 Apr 2005 04:55:10 GMT

Hi,

I would like to know to interpreter this outputs of sar, i would like to known who I want to know when its an bottlenek of cpu, or memory or disk.
In this exemple I have with de cpu:
sar -u
20:43:56 0 6 0 94
20:43:57 0 0 0 100
but i don't know it when the cpu is 100% busy its knows that its bottleneck or not.

Thanks very much!
Carmen.

sar –q

20:13:20 runq-sz %runocc swpq-sz %swpocc
20:13:21 0.0 0 0.0 0
20:15:38 3.0 100 0.0 0
20:15:46 1.0 99 0.0 0

sar -d 1

20:13:20 device %busy avque r+w/s blks/s avwait avserv

20:13:21
sd0 0 0.0 0 0 0.0 0.0
ssd1,e 21 1.7 44 327 0.0 38.5
ssd1,g 0 0.0 0 0 0.0 0.0
ssd2 20 1.4 42 308 0.0 34.2
ssd2,a 0 0.0 0 0 0.0 0.0
ssd2,e 20 1.4 42 308 0.0 34.2

sar –w

SunOS sunz31 5.8 Generic_108528-27 sun4u 04/05/05
20:13:20 swpin/s bswin/s swpot/s bswot/s pswch/s
20:13:21 0.00 0.0 0.00 0.0 1910
20:13:22 0.00 0.0 0.00 0.0 1844
20:13:23 0.00 0.0 0.00 0.0 2699
20:13:24 0.00 0.0 0.00 0.0 325
20:13:25 0.00 0.0 0.00 0.0 492
20:13:26 0.00 0.0 0.00 0.0 3095
20:13:27 0.00 0.0 0.00 0.0 1975

vmstat

procs memory page disk faults cpu
r b w swap free re mf pi po fr de sr s0 s3 s3 s3 in sy cs us sy id
0 0 0 7375256 353360 116 4385 32 0 0 0 0 0 0 0 0 528 4232 1281 1 7 91
0 0 0 7373608 352840 1011 33658 1304 0 0 0 0 0 0 3 0 2754 13015 6377 16 51 33
1 0 0 7360504 340600 1866 17277 24 0 0 0 0 0 0 0 0 546 26799 2939 12 27 61
0 0 0 7372912 351984 2056 28632 184 0 0 0 0 0 0 0 0 2063 27857 5188 19 44 38

sar -u

20:35:10 %usr %sys %wio %idle
20:35:11 10 29 0 61
20:43:49 8 18 2 72
20:43:50 5 22 0 73
20:43:51 1 10 0 89
20:43:52 1 1 0 98
20:43:53 4 15 0 81
20:43:54 7 21 0 72
20:43:55 7 20 1 72
20:43:56 0 6 0 94
20:43:57 0 0 0 100
20:43:58 7 20 4 70
20:43:59 7 25 0 68
20:44:00 8 18 1 73
20:44:01 0 4 0 96
20:44:02 12 3 0

Re: performance with sar

Kent Ostby — Wed, 06 Apr 2005 04:58:43 GMT

document A3949710 explains some of these things.

here's a snippet:

An example sar(1M) output may be similar to the following:

# sar 4 5

17:08:59 %usr %sys %wio %idle
17:09:03 0 0 0 100
17:09:07 0 3 26 71
17:09:11 0 11 89 0
17:09:15 0 3 40 56
17:09:19 0 0 0 100

Average 0 3 31 66

The %usr is the percentage of time the CPU spent running in user code.

The %sys is the percentage of time the CPU spent running system code.

The %wio is the percentage of time the CPU spent blocked on IO. This
only includes block IO, raw IO, or Virtual Memory operations like
pagein/swapin. It does not include time spent on tape IO or terminal
IO. If %wio is high, it could be an indication of a disk bottleneck
or it could simply mean that the CPU has nothing else to do while it
waits for a process to complete its IOs.

The %idle is the percentage of time the CPU is totally idle, with
essentially nothing running or nothing performing disk or other
block/raw IO.

In the above example, we started off with an idle system (%idle=100%).
Then the cp(1) command was issued to make a copy of the kernel. This
caused us to enter system code (%sys), and do disk I/O to copy the
data (%wsio=89). The cp(1) command was constantly reading the source
file then copying to the destination file. So the CPU was either in
the cp(1) code, or it was waiting for I/O to complete. Once the cp was
complete, the system returned to an idle state.

Remember, however, that other processes can make use of the %wsio
time when the CPU is blocked on I/O, thus causing the %wsio to decrease
as %user and $sys increase.

topic performance with sar in Operating System - HP-UX

performance with sar

Re: performance with sar