By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
434,728 Members | 2,480 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 434,728 IT Pros & Developers. It's quick & easy.

UDB V8 FP9a "log data" monitor elements

P: n/a
What is the relationship between the snapshot elements:
Log pages written
Number write log IOs
Number partial page log IOs

There doesn't appear to be any information in my PDF System monitor
guide or the online information center.

When a COMMIT is issued; a log page containing updates must be written
to disk before control is returned to the application program. This
should frequently cause partial page I/O. This log page may be written
multiple times until it fills up, depending on volume of updates/commit.

I'm looking for answers to the following questions:
1. Is partial page I/Os included in write log I/Os?
2. What is counted in Log pages written? Full pages only? Full+partial?

Phil Sherman
Nov 12 '05 #1
Share this Question
Share on Google+
3 Replies


P: n/a
See below. #####

--
Pierre Saint-Jacques
SES Consultants Inc.
514-737-4515
"Phil Sherman" <ps******@ameritech.net> a écrit dans le message de news:
Gh***************@newssvr19.news.prodigy.com...
What is the relationship between the snapshot elements:
Log pages written
Number write log IOs
Number partial page log IOs

There doesn't appear to be any information in my PDF System monitor guide
or the online information center.

When a COMMIT is issued; a log page containing updates must be written to
disk before control is returned to the application program. This should
frequently cause partial page I/O. This log page may be written multiple
times until it fills up, depending on volume of updates/commit.

I'm looking for answers to the following questions:
1. Is partial page I/Os included in write log I/Os? ###### Yes. 2. What is counted in Log pages written? Full pages only? Full+partial? ###### Full and partial.
I believe that if you are the only app. connected and you would run from the
CLP the following:
connect to sample (autocommit on from CLP)
update employee set bonus=bonus+1 where empno='000010' (Ran eight
successive executions)
get snapshot for database on sample
I get:
Log pages written = 8
Log write time (sec.ns) = 0.000000004
Number write log IOs = 8
Number read log IOs = 0
Number partial page log IOs = 8
Number log buffer full = 0

Write Log I/Os and Partials are equal.

HTH, Pierre.
Phil Sherman


Nov 12 '05 #2

P: n/a
Hi Phil:

It is documented but not completely.

Anyway, here is the writeup on "Log Pages Written" element:

log_writes - Number of Log Pages Written monitor element
Description: The number of log pages written to disk by the logger.
Usage: You may use this element with an operating system monitor to
quantify the amount of I/O on a device that is attributable to database
activity.
Note: When log pages are written to disk, the last page might not be
full. In such cases, the partial log page remains in the log buffer,
and additional log records are written to the page. Therefore log pages
might be written to disk by the logger more than once. You should not
use this element to measure the number of pages produced by DB2.
<<<<<<<<<<< the DOC is confusing -- Above it says that partial page
remains in the log buffer, but below is the element which records no.
of partial page log IOs. So db2 does write partial log pages to disk
multiple times till the page is full -- correct ?? >>>>>>>>>>>>>

Number partial page log IOs:
Description: The number of I/O requests issued by the logger for
writing partial log data to the disk.
Usage: Use this element in conjunction with the log_writes,
log_write_time, and num_log_write_io elements to determine if the
current disk is adequate for logging.

Number of write log IOs:
Description: The number of I/O requests issued by the logger for
writing log data to the disk.
Usage: Use this element in conjunction with the log_writes and
log_write_time elements to determine if the current disk is adequate
for logging.
HTH

VIjay

Nov 12 '05 #3

P: n/a
This is exactly what my documentation states. Unfortunately, in
examining it and the names of the data elements, I could make a good
case for the numbers to be either inclusive or exclusive. I just didn't
think of doing what Pierre did and try a simple test - probably because
it was 1am in the morning!

Phil Sherman

UDBDBA wrote:
Hi Phil:

It is documented but not completely.

Anyway, here is the writeup on "Log Pages Written" element:

Nov 12 '05 #4

This discussion thread is closed

Replies have been disabled for this discussion.