• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 2011
  • Last Modified:

Sample size in ANALYZE command

Is there a recommended rule of thumb for choosing what
percentage of a table to analyse ?

I have some large tables (20 million rows) which I want
to analyse on a regular basis.

I don't want to analyse too small a number of rows, as
the stat's probably won't be accurate, and could make
things worse if the wrong execution path is chosen as
a result of inaccurate stat's.

But I don't want to analyse the whole table as this will
take too long.

What percentage will give me a good balance ? 10 ? 20 ?

Does Oracle recommend a suitable figure ?

Any accompanying explanations or URL's would be most
appreciated.

Thanks.
0
myerpz
Asked:
myerpz
1 Solution
 
Mark GeerlingsDatabase AdministratorCommented:
Rather than a percentage, I've seen the number 1024 recommended as the number of rows to count when you want analyze to do a quicker job.  I used that figure in the past when I was working on a system with some multi-million row tables, with good results as far as I could tell.
0
 
myerpzAuthor Commented:
I think the default on the ANALYZE command is 1604.
I just wonder if there's any official word from Oracle
on it, or whether they leave it up to the analyst.
0
 
jammalkCommented:
******************************************************
ANALYZE .... ESTIMATE STATISTICS SAMPLE integer ;

SAMPLE INTEGER:
Specify the amount of data from the analyzed object Oracle should sample to estimate statistics. If you omit this parameter, Oracle samples 1064 rows.
 
The default sample value is adequate for tables up to a few thousand rows. If your tables are larger, specify a higher value for SAMPLE. If you specify more than half of the data, Oracle reads all the data and computes the statistics.
 
ROWS causes Oracle to sample integer rows of the table or cluster or integer entries from the index. The integer must be at least 1.

PERCENT causes Oracle to sample integer percent of the rows from the table or cluster or integer percent of the index entries. The integer can range from 1 to 99.
******************************************************
I think 20% of the rows is a good figure estimate the statistics.
0
Independent Software Vendors: We Want Your Opinion

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

 
Mark GeerlingsDatabase AdministratorCommented:
The others are correct, the number is 1064, not 1024 as I said in my earlier comment.  I don't know what is magical about this number, but unless you know or suspect that the first 1,000 or so records in the table are not representative of the majority of records, it is likely much faster for analyze to do this number of rows than a fixed (even small) percentage of a large table.
0
 
dbrowerCommented:
I have used this script for a number of years an it seems to do the trick. It is base on the size of the table not the number of rows.

REM *******************************************************
REM
REM File: C:\Dba\analytab.SQL
REM
REM ******************************************************
REM  This script will produce and run the script
REM  ANALYTB2.SQL to analyze
REM  all tables and create the listing analytab.rpt
REM
REM ******************************************************

set echo off
set heading off
set pagesize 50000
set feedback off
spool c:\dba\analytb2.sql
SELECT 'ANALYZE TABLE ' || RTRIM( owner ) || '.' ||
          segment_name || ' COMPUTE STATISTICS ;'
     FROM sys.dba_segments
     WHERE segment_type = 'TABLE'
          AND owner NOT IN ('SYS','SYSTEM','VOSER')
          AND bytes < 5000000
UNION
SELECT 'ANALYZE TABLE ' || RTRIM( owner ) || '.' ||
          segment_name || ' ESTIMATE STATISTICS SAMPLE 30 PERCENT;'
     FROM sys.dba_segments
     WHERE segment_type = 'TABLE'
          AND owner NOT IN ('SYS','SYSTEM','VOSER')
          AND bytes between 5000001 AND 30000000
UNION
SELECT 'ANALYZE TABLE ' || RTRIM( owner ) || '.' ||
          segment_name || ' ESTIMATE STATISTICS SAMPLE 10 PERCENT;'
     FROM sys.dba_segments
     WHERE segment_type = 'TABLE'
          AND owner NOT IN ('SYS','SYSTEM','VOSER')
          AND bytes between 30000001 AND 50000000
UNION
SELECT 'ANALYZE TABLE ' || RTRIM( owner ) || '.' ||
          segment_name || ' ESTIMATE STATISTICS SAMPLE 5 PERCENT;'
     FROM sys.dba_segments
     WHERE segment_type = 'TABLE'
          AND owner IN ('SYS','SYSTEM')
          AND bytes > 50000000
ORDER BY 1
;

spool off

set echo on
set feedback on

@C:\dba\analytb2

0
 
chdbaCommented:
I'm not experienced with analyze schema.

But dbrower, are you sure, that you always used THIS script for a number of years?

What if a table is > 50000000 Bytes?

Ooops, is there perhaps a "NOT" missing ;-)?

Good luck.
0
 
orbajCommented:
Hi

Why not test different samples percentage for both time and statistics accuracy ?

 
0

Featured Post

Nothing ever in the clear!

This technical paper will help you implement VMware’s VM encryption as well as implement Veeam encryption which together will achieve the nothing ever in the clear goal. If a bad guy steals VMs, backups or traffic they get nothing.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now