Architecture
  Internals
  Backup/Recovery
  Security
  Tuning
  Quick Commands
  Packages
  PL/SQL History

 

 

Unix Tuning Guide for DBA's
  1. The DBA must resist the temptation to tune what is measurable rather than what is important.
  2. Resist the temptation to tune just because you can.
  3. Do not jump to conclusions and keep an open mind. Use your past experiences but do not let them limit you.
  4. Be persistant.
Monitoring CPU
Command Returns
uptime Rough estimate of system load.
ps -ef Information of the currently running processes.
ps -ef|wc -l count of total processess.
pstree Heirarchical display of parent and child process
top Most active processes that are continuously updated.
   
vmstat interval [count]

us: Percentage of CPU cycles spent as user time (applications)

sy: Percentage of CPU cycles spent as system time (I/O requests, other general operating system services). This column can be high if your system is experiencing paging and swapping.

id: Percentage of CPU that went used during the interval, Idle time.

 

Monitoring Memory
Command Returns
vmstat interval [count]

r: number of runnable processes

b
: number of blocked processes (waiting on I/O)

w
: number of swapped out runnable processes


avm
: Number of active virtual memory pages

fre
: Number of memory pages on the free list. Paging activity is identified by significant decreases in this column.

re
: Number of page reclaims: pages placed on the free list but reclaimed before the page was reused.

pi: Number of pages paged-in. You will see this column spike when new processes are started, but should quickly level off.

po: Number of pages paged out. This is the most important column for identifing paging. This should ideally be zero or close to zero. Anything greater then zero indicates that processes are contending for the available memory and the system is paging.

fr: Pages freed by the virutal memory facility during interval.

Note: The number should be 0 in the w column, otherwise it is a indication of a serious memory shortage.

 

Monitoring I/O
Command Options
iostat

-c: Repeat the display count times. Unless the -I flag is in effect, the first display is for the time since a reboot and each subse-quent report is for the time period since the last display. If no wait interval is specified, the default is 1 second.

-C: Show cpu statistics. This is enabled by default unless the -d, -D, or -T flags are used.

-d: Show disk statistics. This is the default. Displays kilobytes per transfer, number of transfers, and megabytes transfered. Use of this flag disables display of cpu and tty statistics.

-D: Show alternate disk statistics. Displays kilobytes transfered,
number of transfers, and time spent in transfers. Use of this flag disables the default display.

-I: Show the running total values, rather than an average.

-M: Extract values associated with the name list from the specified
core instead of the default ``/dev/mem''.

-N: Extract the name list from the specified system instead of the de-
fault ``/bsd''.

-T: Show tty statistics. This is enabled by default unless the -C, -d, or -D flags are used.

-w: Pause wait seconds between each display. If no repeat count is specified, the default is infinity.

 

About Oraclenotes.com  |  Advertise  |  Contribute   |  Disclaimer 
© Copyright 1999 - 2003 Oraclenotes.com. All Rights Reserved.
Use of this website signifies your agreement to the Terms of Use.

Not affliated or endorced by Oracle Corporation.