Oracle-L: RE: OT : Basic Critical O.S. Values that Trigger Problem Alert

From: DENNIS WILLIAMS <DWILLIAMS_at_LIFETOUCH.COM>
Date: Tue, 30 Jul 2002 05:33:19 -0800
Message-ID: <F001.004A5A40.20020730053319@fatcity.com>

Vivek

Here is a search string that should work. The list server will probably break it into two pieces, so you'll have to patch it back together. Good luck on your project.

http://search.barnesandnoble.com/booksearch/isbnInquiry.asp?userid=1G60ZMKA1 J&isbn=007222360X

Dennis Williams
DBA
Lifetouch, Inc.
dwilliams_at_lifetouch.com

-----Original Message-----

Sent: Monday, July 29, 2002 11:54 PM
To: DENNIS WILLIAMS
Cc: ORACLE-L_at_fatcity.com

Book "Oracle High-Performance Tuning with STATSPACK" by Don Burleson

What is the PUBLISHER / Any Other Details ?

Thanks indeed Dennis .

-----Original Message-----

Sent: Tuesday, July 30, 2002 3:24 AM
To: Multiple recipients of list ORACLE-L

Vivek - This sort of thing must be used with some intelligence. It is difficult to provide guidelines that work for all sites. Trending is very important. If a lightly loaded site that suddenly experiences high statistics, that may something that needs reviewed. Another site may always be highly loaded so the statistics may always look abysmal compared to the lightly loaded site. Time of day is also important. Ideally you collect statistics 24x7 and compare the trends to user events. And see how the statistics change over time. Trends are much more important than single statistics out of context. Don Burleson devotes several chapters to operating system indicators and trending in his book "Oracle High-Performance Tuning with STATSPACK".

In the end, the critical measurement is user response time. If the users think performance stinks, then by definition it stinks. Perception is reality. Users don't care if the problem is with the network, the server, the Web server, the application server, or the database. If you have an opportunity to configure a test that simulates what the users see, that is ideal. Then when the users say that performance is bad, then you can show them your trend line. Those sort of facts saves a lot of argument. If your trend shows performance was bad, you will be inclined to check it out. Otherwise, the user may be convinced that it was their perception. Dennis Williams
DBA
Lifetouch, Inc.
dwilliams_at_lifetouch.com

-----Original Message-----

Sent: Monday, July 29, 2002 2:48 AM
To: Multiple recipients of list ORACLE-L

We are Trying to make a General Document to be forwarded to Customers which should allow them to know when they are performing far below normal

At the Operating System Level we are trying to Identify Practical Critical Values which when below respective Threshold Limits which would give the alert about a potential problem .

We are Looking for these in Areas of :-

Network Thruput
Memory Utilization
Swap Utilization
IO Utilization

Would apreciate actual Commands used (preferably those Generic across different O.S.) & respective Critical Threshold Limit Values for the Above

EXAMPLE For Network thruput Between APPLICATION Server machine & Database Server Machine
what , by experience , are the parameters & their respective Minimum threshold Values which would let us know that there is a Severe problem therein ?

NOTE - We have generally been measuring this by Manually ftping a Big File , about 100MB , between APP & DB Server machines , noting the thruput Displayed in (kbytes/s) on Completion & Converting this Value to Mega Bits / Second
(i.e. MBPS) . If this Value is Less than 40MBPS for a 100 MBPS Cable we know there is a PRoblem with Network Bandwidth.

Miscellaneous - Some Threshold Limits known to us :-

Command - vmstat 5 3
Virtual Memory Statistics: (pagesize = 8192)

  procs      memory        pages                            intr

cpu
  r w u act free wire fault cow zero react pin pout in sy cs us sy id
  3 1K 34 266K 84K 32K 811M 132M 339M 635 193M 0 188 28K 1K 16 7 77
  3 1K 33 267K 84K 32K 410 71 151 0 131 0 494 2K 4K 4 2 93
  3 1K 36 269K 82K 32K 5459 1720 807 0 3016 0 471 3K 4K 37 5 58

Utilization of CPU due to Operating System (Internal) Operations (%sy) Exceeding Utilization due to user Applications (%us)
Average Wait of CPU for IO to Complete (%wio) Greater than (>) 30 % [ From sar Command ]
Utilization of CPU due to Operating System (Internal) Operations (%sy) > 30 %
CPU Utilization - If Total CPU Utilization Consistently Near 0% Idle Or further Coupled with any of the following :- a)Abnormally High Wait for IO ( > 30 %) [ From sar Command ] b)Abnormally High Operating System CPU Utilization ( > 30 %) c)Abnormally High Run Queue ["r" > (3 * Number of CPUs)]

THANKS

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.com
-- 
Author: DENNIS WILLIAMS
  INET: DWILLIAMS_at_LIFETOUCH.COM

Fat City Network Services    -- (858) 538-5051  FAX: (858) 538-5051
San Diego, California        -- Public Internet access / Mailing Lists


--------------------------------------------------------------------

To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Received on Tue Jul 30 2002 - 08:33:19 CDT