system tuning and analysis:

The areas you'll want to review and monitor on your database server using Unix utilities are:

• CPU performance. How busy are the CPUs? How big is the backlog of processes waiting to run on your CPUs?
• Memory usage. How much memory is being used? Are you swapping out to disk because too much is being used? Do you have unused memory that would allow you to increase the size of your database buffers?
• Disk output. What is the disk I/O throughput? Is a particular disk a bottleneck?
• Network usage. What's the network utilization? Are there network collisions and errors that may be slowing down your database applications?

sar, for CPU, disk, and memory statistics vmstat, for CPU and virtual memory statistics mpstat, for per-CPU statistics iostat, for disk I/O throughput statistics ps, for Unix processes statistics top, for top Unix processes statistics netstat, for network statistics uptime, for the system uptime and the average load on the system. [@more@]

1. check total physical memory:

/usr/sbin/prtconf | grep Memory

2. check total CPU:

/usr/sbin/psrinfo | grep Statistics | wc -l

/usr/platform/`uname -i`/sbin/prtdiag -v

3. check memory usage

ps -el

SZ memory usage in 4096bytes

#check very long process name:
/usr/ucb/ps -auxww

prstat -s size

sorted memory usage

4.vmstat

po, pi page in and out
sr > 200 (too bad)

5. sar

-a

-u

%WIO pecent time cpu wait for I/O
add usr%+sys%
<30% not bad
>30% <60% so so
>60% poor

-p

paging statistics

-q

runq -sz number of processes ready to run now

6. uptime
<2 per cpu excellent
2-4 per cpu fair
>4 per cpu poor

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/67/viewspace-967763/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/67/viewspace-967763/