Comments on: Oracle 11g: Incremental Global Statistics On Partitioned Tables http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/?utm_source=rss&utm_medium=rss&utm_campaign=oracle-11g-incremental-global-statistics-on-partitioned-tables Data, Databases, Performance & Scalability Mon, 30 Jan 2012 17:05:12 -0500 hourly 1 http://wordpress.org/?v=3.3.1 By: Greg Rahn http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-23991 Greg Rahn Thu, 27 Oct 2011 00:59:29 +0000 http://structureddata.org/?p=70#comment-23991 I only have a base 11.1 setup and I have not tried it. I only have a base 11.1 setup and I have not tried it.

]]> By: DBA http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-23976 DBA Wed, 26 Oct 2011 14:09:13 +0000 http://structureddata.org/?p=70#comment-23976 Greg, did you have a chance to test my small test case on 11.1? We applied SP9 to our 11.1.0.7 version and I am still not seeing the incremental stats work as expected. Greg, did you have a chance to test my small test case on 11.1? We applied SP9 to our 11.1.0.7 version and I am still not seeing the incremental stats work as expected.

]]> By: DBA http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-17897 DBA Fri, 20 May 2011 21:08:58 +0000 http://structureddata.org/?p=70#comment-17897 ghassan, I did not try that, but granularity by default is AUTO. Anyway it's good to know that it is fixed in 11.2.0.2. ghassan,
I did not try that, but granularity by default is AUTO. Anyway it’s good to know that it is fixed in 11.2.0.2.

]]> By: ghassan http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-17757 ghassan Fri, 13 May 2011 16:34:26 +0000 http://structureddata.org/?p=70#comment-17757 DBA, I tried your testcase on an 11.2.0.2 (don't have access to 11.1), and it does not reproduce. on your version, did you try on an 11.1, without specifying the 'granularity'? DBA,
I tried your testcase on an 11.2.0.2 (don’t have access to 11.1), and it does not reproduce.
on your version, did you try on an 11.1, without specifying the ‘granularity’?

]]> By: Greg Rahn http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-17300 Greg Rahn Mon, 11 Apr 2011 18:09:38 +0000 http://structureddata.org/?p=70#comment-17300 No worries - it's <strong>[sourcecode][/sourcecode]</strong>. I've added it. I'll look at this when I have a spare moment and get back. Stay tuned! No worries – it’s . I’ve added it. I’ll look at this when I have a spare moment and get back. Stay tuned!

]]> By: DBA http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-17299 DBA Mon, 11 Apr 2011 16:26:00 +0000 http://structureddata.org/?p=70#comment-17299 Sorry about the formatting. I tried {code} tag but it didn't work. Sorry about the formatting. I tried {code} tag but it didn’t work.

]]> By: DBA http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-17298 DBA Mon, 11 Apr 2011 16:25:06 +0000 http://structureddata.org/?p=70#comment-17298 Here is the test case. Watch for the last_analyzed column. [sourcecode] SQL> ALTER SESSION SET nls_date_format='DD-MON-YYYY HH12:MI:SS AM'; Session altered. SQL> SQL> DROP TABLE t PURGE; Table dropped. SQL> prompt ==>Create a partitioned table T using interval partitioning ==>Create a partitioned table T using interval partitioning SQL> SQL> CREATE TABLE t ( 2 id NUMBER CONSTRAINT pk_t PRIMARY KEY, 3 sample_time DATE, 4 description VARCHAR2 (100) 5 ) 6 PARTITION BY RANGE (sample_time) 7 INTERVAL ( NUMTODSINTERVAL ('1', 8 'DAY') ) 9 (PARTITION p1 10 VALUES LESS THAN 11 (TO_DATE ('2011-04-01', 12 'YYYY-MM-DD'))); Table created. SQL> SQL> prompt ==>Check the preferences for the table ==>Check the preferences for the table SQL> SQL> SELECT table_name, 2 DBMS_STATS.get_prefs (pname => 'INCREMENTAL', 3 ownname => USER, 4 tabname => table_name) 5 incremental, 6 DBMS_STATS.get_prefs (pname => 'GRANULARITY', 7 ownname => USER, 8 tabname => table_name) 9 granularity, 10 DBMS_STATS.get_prefs (pname => 'PUBLISH', 11 ownname => USER, 12 tabname => table_name) 13 publish, 14 DBMS_STATS.get_prefs (pname => 'ESTIMATE_PERCENT', 15 ownname => USER, 16 tabname => table_name) 17 estimate_percent 18 FROM user_tables 19 WHERE table_name = 'T' 20 ORDER BY table_name; TABLE_NAME|INCREMENTAL |GRANULARITY |PUBLISH |ESTIMATE_PERCENT ----------|------------|------------|------------|------------------------------ T |FALSE |AUTO |TRUE |DBMS_STATS.AUTO_SAMPLE_SIZE SQL> SQL> prompt ==>Set table to use incremental stats ==>Set table to use incremental stats SQL> SQL> BEGIN 2 DBMS_STATS.set_table_prefs (ownname => USER, 3 tabname => 'T', 4 pname => 'INCREMENTAL', 5 pvalue => 'TRUE'); 6 END; 7 / PL/SQL procedure successfully completed. SQL> SQL> prompt ==>Check the preferences for the table ==>Check the preferences for the table SQL> SQL> SELECT table_name, 2 DBMS_STATS.get_prefs (pname => 'INCREMENTAL', 3 ownname => USER, 4 tabname => table_name) 5 incremental, 6 DBMS_STATS.get_prefs (pname => 'GRANULARITY', 7 ownname => USER, 8 tabname => table_name) 9 granularity, 10 DBMS_STATS.get_prefs (pname => 'PUBLISH', 11 ownname => USER, 12 tabname => table_name) 13 publish, 14 DBMS_STATS.get_prefs (pname => 'ESTIMATE_PERCENT', 15 ownname => USER, 16 tabname => table_name) 17 estimate_percent 18 FROM user_tables 19 WHERE table_name = 'T' 20 ORDER BY table_name; TABLE_NAME|INCREMENTAL |GRANULARITY |PUBLISH |ESTIMATE_PERCENT ----------|------------|------------|------------|------------------------------ T |TRUE |AUTO |TRUE |DBMS_STATS.AUTO_SAMPLE_SIZE SQL> SQL> prompt ==>Insert 2 rows ==>Insert 2 rows SQL> SQL> INSERT INTO t 2 VALUES (1, SYSDATE - 5, 'A'); 1 row created. SQL> SQL> INSERT INTO t 2 VALUES (2, SYSDATE - 4, 'A'); 1 row created. SQL> SQL> COMMIT; Commit complete. SQL> SQL> SELECT * FROM t; ID|SAMPLE_TIME |DESCRIPTION ----------|-----------------------|------------ 1|06-APR-2011 12:21:31 PM|A 2|07-APR-2011 12:21:31 PM|A SQL> SQL> BEGIN 2 DBMS_STATS.gather_table_stats (ownname => USER, 3 tabname => 'T', 4 estimate_percent => DBMS_STATS.auto_sample_size, 5 method_opt => 'FOR ALL COLUMNS SIZE 1', 6 granularity => 'AUTO', 7 degree => 2, 8 cascade => TRUE, 9 no_invalidate => TRUE); 10 END; 11 / PL/SQL procedure successfully completed. SQL> SQL> SELECT table_name, num_rows, last_analyzed 2 FROM user_tables 3 WHERE table_name = 'T'; TABLE_NAME| NUM_ROWS|LAST_ANALYZED ----------|----------|----------------------- T | 2|11-APR-2011 12:21:32 PM SQL> SQL> SELECT table_name, 2 partition_name, 3 partition_position, 4 num_rows, 5 last_analyzed 6 FROM user_tab_partitions 7 WHERE table_name = 'T' 8 ORDER BY partition_position; TABLE_NAME|PARTITION_NAME|PARTITION_POSITION| NUM_ROWS|LAST_ANALYZED ----------|--------------|------------------|----------|----------------------- T |P1 | 1| 0|11-APR-2011 12:21:31 PM T |SYS_P123420 | 2| 1|11-APR-2011 12:21:31 PM T |SYS_P123421 | 3| 1|11-APR-2011 12:21:31 PM SQL> SQL> SELECT index_name, num_rows, last_analyzed 2 FROM user_indexes 3 WHERE table_name = 'T'; INDEX_NAME| NUM_ROWS|LAST_ANALYZED ----------|----------|----------------------- PK_T | 2|11-APR-2011 12:21:32 PM SQL> SQL> EXEC dbms_lock.sleep(2); PL/SQL procedure successfully completed. SQL> SQL> prompt ==>Insert 2 rows ==>Insert 2 rows SQL> SQL> INSERT INTO t 2 VALUES (3, SYSDATE - 3, 'A'); 1 row created. SQL> SQL> INSERT INTO t 2 VALUES (4, SYSDATE - 2, 'A'); 1 row created. SQL> SQL> COMMIT; Commit complete. SQL> SQL> SELECT * FROM t; ID|SAMPLE_TIME |DESCRIPTION ----------|-----------------------|------------ 1|06-APR-2011 12:21:31 PM|A 2|07-APR-2011 12:21:31 PM|A 3|08-APR-2011 12:21:34 PM|A 4|09-APR-2011 12:21:34 PM|A SQL> SQL> BEGIN 2 DBMS_STATS.gather_table_stats (ownname => USER, 3 tabname => 'T', 4 estimate_percent => DBMS_STATS.auto_sample_size, 5 method_opt => 'FOR ALL COLUMNS SIZE 1', 6 granularity => 'AUTO', 7 degree => 2, 8 cascade => TRUE, 9 no_invalidate => TRUE); 10 END; 11 / PL/SQL procedure successfully completed. SQL> SQL> SELECT table_name, num_rows, last_analyzed 2 FROM user_tables 3 WHERE table_name = 'T'; TABLE_NAME| NUM_ROWS|LAST_ANALYZED ----------|----------|----------------------- T | 4|11-APR-2011 12:21:35 PM SQL> SQL> SELECT table_name, 2 partition_name, 3 partition_position, 4 num_rows, 5 last_analyzed 6 FROM user_tab_partitions 7 WHERE table_name = 'T' 8 ORDER BY partition_position; TABLE_NAME|PARTITION_NAME|PARTITION_POSITION| NUM_ROWS|LAST_ANALYZED ----------|--------------|------------------|----------|----------------------- T |P1 | 1| 0|11-APR-2011 12:21:31 PM T |SYS_P123420 | 2| 1|11-APR-2011 12:21:34 PM T |SYS_P123421 | 3| 1|11-APR-2011 12:21:34 PM T |SYS_P123422 | 4| 1|11-APR-2011 12:21:34 PM T |SYS_P123423 | 5| 1|11-APR-2011 12:21:34 PM SQL> SQL> SELECT index_name, num_rows, last_analyzed 2 FROM user_indexes 3 WHERE table_name = 'T'; INDEX_NAME| NUM_ROWS|LAST_ANALYZED ----------|----------|----------------------- PK_T | 4|11-APR-2011 12:21:35 PM SQL> SQL> EXEC dbms_lock.sleep(2); PL/SQL procedure successfully completed. SQL> SQL> prompt ==>No insert here, just gather stats ==>No insert here, just gather stats SQL> SQL> BEGIN 2 DBMS_STATS.gather_table_stats (ownname => USER, 3 tabname => 'T', 4 estimate_percent => DBMS_STATS.auto_sample_size, 5 method_opt => 'FOR ALL COLUMNS SIZE 1', 6 granularity => 'AUTO', 7 degree => 2, 8 cascade => TRUE, 9 no_invalidate => TRUE); 10 END; 11 / PL/SQL procedure successfully completed. SQL> SQL> SELECT table_name, num_rows, last_analyzed 2 FROM user_tables 3 WHERE table_name = 'T'; TABLE_NAME| NUM_ROWS|LAST_ANALYZED ----------|----------|----------------------- T | 4|11-APR-2011 12:21:38 PM SQL> SQL> SELECT table_name, 2 partition_name, 3 partition_position, 4 num_rows, 5 last_analyzed 6 FROM user_tab_partitions 7 WHERE table_name = 'T' 8 ORDER BY partition_position; TABLE_NAME|PARTITION_NAME|PARTITION_POSITION| NUM_ROWS|LAST_ANALYZED ----------|--------------|------------------|----------|----------------------- T |P1 | 1| 0|11-APR-2011 12:21:31 PM T |SYS_P123420 | 2| 1|11-APR-2011 12:21:37 PM T |SYS_P123421 | 3| 1|11-APR-2011 12:21:37 PM T |SYS_P123422 | 4| 1|11-APR-2011 12:21:37 PM T |SYS_P123423 | 5| 1|11-APR-2011 12:21:37 PM SQL> SQL> SELECT index_name, num_rows, last_analyzed 2 FROM user_indexes 3 WHERE table_name = 'T'; INDEX_NAME| NUM_ROWS|LAST_ANALYZED ----------|----------|----------------------- PK_T | 4|11-APR-2011 12:21:38 PM SQL> SQL> EXEC dbms_lock.sleep(2); PL/SQL procedure successfully completed. SQL> SQL> prompt ==>No insert here, just gather stats ==>No insert here, just gather stats SQL> SQL> BEGIN 2 DBMS_STATS.gather_table_stats (ownname => USER, 3 tabname => 'T', 4 estimate_percent => DBMS_STATS.auto_sample_size, 5 method_opt => 'FOR ALL COLUMNS SIZE 1', 6 granularity => 'AUTO', 7 degree => 2, 8 cascade => TRUE, 9 no_invalidate => TRUE); 10 END; 11 / PL/SQL procedure successfully completed. SQL> SQL> SELECT table_name, num_rows, last_analyzed 2 FROM user_tables 3 WHERE table_name = 'T'; TABLE_NAME| NUM_ROWS|LAST_ANALYZED ----------|----------|----------------------- T | 4|11-APR-2011 12:21:41 PM SQL> SQL> SELECT table_name, 2 partition_name, 3 partition_position, 4 num_rows, 5 last_analyzed 6 FROM user_tab_partitions 7 WHERE table_name = 'T' 8 ORDER BY partition_position; TABLE_NAME|PARTITION_NAME|PARTITION_POSITION| NUM_ROWS|LAST_ANALYZED ----------|--------------|------------------|----------|----------------------- T |P1 | 1| 0|11-APR-2011 12:21:31 PM T |SYS_P123420 | 2| 1|11-APR-2011 12:21:41 PM T |SYS_P123421 | 3| 1|11-APR-2011 12:21:41 PM T |SYS_P123422 | 4| 1|11-APR-2011 12:21:41 PM T |SYS_P123423 | 5| 1|11-APR-2011 12:21:41 PM SQL> SQL> SELECT index_name, num_rows, last_analyzed 2 FROM user_indexes 3 WHERE table_name = 'T'; INDEX_NAME| NUM_ROWS|LAST_ANALYZED ----------|----------|----------------------- PK_T | 4|11-APR-2011 12:21:41 PM SQL> SQL> exit [/sourcecode] Here is the test case. Watch for the last_analyzed column.


SQL> ALTER SESSION SET nls_date_format='DD-MON-YYYY HH12:MI:SS AM';

Session altered.

SQL>
SQL> DROP TABLE t PURGE;

Table dropped.

SQL> prompt ==>Create a partitioned table T using interval partitioning
==>Create a partitioned table T using interval partitioning
SQL>
SQL> CREATE TABLE t (
  2  id NUMBER CONSTRAINT pk_t PRIMARY KEY,
  3  sample_time DATE,
  4  description VARCHAR2 (100)
  5  )
  6  PARTITION BY RANGE (sample_time)
  7    INTERVAL ( NUMTODSINTERVAL ('1',
  8  				   'DAY') )
  9    (PARTITION p1
 10  	  VALUES LESS THAN
 11  	    (TO_DATE ('2011-04-01',
 12  		      'YYYY-MM-DD')));

Table created.

SQL>
SQL> prompt ==>Check the preferences for the table
==>Check the preferences for the table
SQL>
SQL>   SELECT table_name,
  2  	      DBMS_STATS.get_prefs (pname   => 'INCREMENTAL',
  3  				    ownname => USER,
  4  				    tabname => table_name)
  5  		incremental,
  6  	      DBMS_STATS.get_prefs (pname   => 'GRANULARITY',
  7  				    ownname => USER,
  8  				    tabname => table_name)
  9  		granularity,
 10  	      DBMS_STATS.get_prefs (pname   => 'PUBLISH',
 11  				    ownname => USER,
 12  				    tabname => table_name)
 13  		publish,
 14  	      DBMS_STATS.get_prefs (pname   => 'ESTIMATE_PERCENT',
 15  				    ownname => USER,
 16  				    tabname => table_name)
 17  		estimate_percent
 18  	 FROM user_tables
 19  	WHERE table_name = 'T'
 20  ORDER BY table_name;

TABLE_NAME|INCREMENTAL |GRANULARITY |PUBLISH     |ESTIMATE_PERCENT
----------|------------|------------|------------|------------------------------
T         |FALSE       |AUTO        |TRUE        |DBMS_STATS.AUTO_SAMPLE_SIZE

SQL>
SQL> prompt ==>Set table to use incremental stats
==>Set table to use incremental stats
SQL>
SQL> BEGIN
  2    DBMS_STATS.set_table_prefs (ownname => USER,
  3  				   tabname => 'T',
  4  				   pname   => 'INCREMENTAL',
  5  				   pvalue  => 'TRUE');
  6  END;
  7  /

PL/SQL procedure successfully completed.

SQL>
SQL> prompt ==>Check the preferences for the table
==>Check the preferences for the table
SQL>
SQL>   SELECT table_name,
  2  	      DBMS_STATS.get_prefs (pname   => 'INCREMENTAL',
  3  				    ownname => USER,
  4  				    tabname => table_name)
  5  		incremental,
  6  	      DBMS_STATS.get_prefs (pname   => 'GRANULARITY',
  7  				    ownname => USER,
  8  				    tabname => table_name)
  9  		granularity,
 10  	      DBMS_STATS.get_prefs (pname   => 'PUBLISH',
 11  				    ownname => USER,
 12  				    tabname => table_name)
 13  		publish,
 14  	      DBMS_STATS.get_prefs (pname   => 'ESTIMATE_PERCENT',
 15  				    ownname => USER,
 16  				    tabname => table_name)
 17  		estimate_percent
 18  	 FROM user_tables
 19  	WHERE table_name = 'T'
 20  ORDER BY table_name;

TABLE_NAME|INCREMENTAL |GRANULARITY |PUBLISH     |ESTIMATE_PERCENT
----------|------------|------------|------------|------------------------------
T         |TRUE        |AUTO        |TRUE        |DBMS_STATS.AUTO_SAMPLE_SIZE

SQL>
SQL> prompt ==>Insert 2 rows
==>Insert 2 rows
SQL>
SQL> INSERT INTO t
  2  	  VALUES (1, SYSDATE - 5, 'A');

1 row created.

SQL>
SQL> INSERT INTO t
  2  	  VALUES (2, SYSDATE - 4, 'A');

1 row created.

SQL>
SQL> COMMIT;

Commit complete.

SQL>
SQL> SELECT * FROM t;

        ID|SAMPLE_TIME            |DESCRIPTION
----------|-----------------------|------------
         1|06-APR-2011 12:21:31 PM|A
         2|07-APR-2011 12:21:31 PM|A

SQL>
SQL> BEGIN
  2    DBMS_STATS.gather_table_stats (ownname	       => USER,
  3  				      tabname	       => 'T',
  4  				      estimate_percent => DBMS_STATS.auto_sample_size,
  5  				      method_opt       => 'FOR ALL COLUMNS SIZE 1',
  6  				      granularity      => 'AUTO',
  7  				      degree	       => 2,
  8  				      cascade	       => TRUE,
  9  				      no_invalidate    => TRUE);
 10  END;
 11  /

PL/SQL procedure successfully completed.

SQL>
SQL> SELECT table_name, num_rows, last_analyzed
  2    FROM user_tables
  3   WHERE table_name = 'T';

TABLE_NAME|  NUM_ROWS|LAST_ANALYZED
----------|----------|-----------------------
T         |         2|11-APR-2011 12:21:32 PM

SQL>
SQL>   SELECT table_name,
  2  	      partition_name,
  3  	      partition_position,
  4  	      num_rows,
  5  	      last_analyzed
  6  	 FROM user_tab_partitions
  7  	WHERE table_name = 'T'
  8  ORDER BY partition_position;

TABLE_NAME|PARTITION_NAME|PARTITION_POSITION|  NUM_ROWS|LAST_ANALYZED
----------|--------------|------------------|----------|-----------------------
T         |P1            |                 1|         0|11-APR-2011 12:21:31 PM
T         |SYS_P123420   |                 2|         1|11-APR-2011 12:21:31 PM
T         |SYS_P123421   |                 3|         1|11-APR-2011 12:21:31 PM

SQL>
SQL> SELECT index_name, num_rows, last_analyzed
  2    FROM user_indexes
  3   WHERE table_name = 'T';

INDEX_NAME|  NUM_ROWS|LAST_ANALYZED
----------|----------|-----------------------
PK_T      |         2|11-APR-2011 12:21:32 PM

SQL>
SQL> EXEC dbms_lock.sleep(2);

PL/SQL procedure successfully completed.

SQL>
SQL> prompt ==>Insert 2 rows
==>Insert 2 rows
SQL>
SQL> INSERT INTO t
  2  	  VALUES (3, SYSDATE - 3, 'A');

1 row created.

SQL>
SQL> INSERT INTO t
  2  	  VALUES (4, SYSDATE - 2, 'A');

1 row created.

SQL>
SQL> COMMIT;

Commit complete.

SQL>
SQL> SELECT * FROM t;

        ID|SAMPLE_TIME            |DESCRIPTION
----------|-----------------------|------------
         1|06-APR-2011 12:21:31 PM|A
         2|07-APR-2011 12:21:31 PM|A
         3|08-APR-2011 12:21:34 PM|A
         4|09-APR-2011 12:21:34 PM|A

SQL>
SQL> BEGIN
  2    DBMS_STATS.gather_table_stats (ownname	       => USER,
  3  				      tabname	       => 'T',
  4  				      estimate_percent => DBMS_STATS.auto_sample_size,
  5  				      method_opt       => 'FOR ALL COLUMNS SIZE 1',
  6  				      granularity      => 'AUTO',
  7  				      degree	       => 2,
  8  				      cascade	       => TRUE,
  9  				      no_invalidate    => TRUE);
 10  END;
 11  /

PL/SQL procedure successfully completed.

SQL>
SQL> SELECT table_name, num_rows, last_analyzed
  2    FROM user_tables
  3   WHERE table_name = 'T';

TABLE_NAME|  NUM_ROWS|LAST_ANALYZED
----------|----------|-----------------------
T         |         4|11-APR-2011 12:21:35 PM

SQL>
SQL>   SELECT table_name,
  2  	      partition_name,
  3  	      partition_position,
  4  	      num_rows,
  5  	      last_analyzed
  6  	 FROM user_tab_partitions
  7  	WHERE table_name = 'T'
  8  ORDER BY partition_position;

TABLE_NAME|PARTITION_NAME|PARTITION_POSITION|  NUM_ROWS|LAST_ANALYZED
----------|--------------|------------------|----------|-----------------------
T         |P1            |                 1|         0|11-APR-2011 12:21:31 PM
T         |SYS_P123420   |                 2|         1|11-APR-2011 12:21:34 PM
T         |SYS_P123421   |                 3|         1|11-APR-2011 12:21:34 PM
T         |SYS_P123422   |                 4|         1|11-APR-2011 12:21:34 PM
T         |SYS_P123423   |                 5|         1|11-APR-2011 12:21:34 PM

SQL>
SQL> SELECT index_name, num_rows, last_analyzed
  2    FROM user_indexes
  3   WHERE table_name = 'T';

INDEX_NAME|  NUM_ROWS|LAST_ANALYZED
----------|----------|-----------------------
PK_T      |         4|11-APR-2011 12:21:35 PM

SQL>
SQL> EXEC dbms_lock.sleep(2);

PL/SQL procedure successfully completed.

SQL>
SQL> prompt ==>No insert here, just gather stats
==>No insert here, just gather stats
SQL>
SQL> BEGIN
  2    DBMS_STATS.gather_table_stats (ownname	       => USER,
  3  				      tabname	       => 'T',
  4  				      estimate_percent => DBMS_STATS.auto_sample_size,
  5  				      method_opt       => 'FOR ALL COLUMNS SIZE 1',
  6  				      granularity      => 'AUTO',
  7  				      degree	       => 2,
  8  				      cascade	       => TRUE,
  9  				      no_invalidate    => TRUE);
 10  END;
 11  /

PL/SQL procedure successfully completed.

SQL>
SQL> SELECT table_name, num_rows, last_analyzed
  2    FROM user_tables
  3   WHERE table_name = 'T';

TABLE_NAME|  NUM_ROWS|LAST_ANALYZED
----------|----------|-----------------------
T         |         4|11-APR-2011 12:21:38 PM

SQL>
SQL>   SELECT table_name,
  2  	      partition_name,
  3  	      partition_position,
  4  	      num_rows,
  5  	      last_analyzed
  6  	 FROM user_tab_partitions
  7  	WHERE table_name = 'T'
  8  ORDER BY partition_position;

TABLE_NAME|PARTITION_NAME|PARTITION_POSITION|  NUM_ROWS|LAST_ANALYZED
----------|--------------|------------------|----------|-----------------------
T         |P1            |                 1|         0|11-APR-2011 12:21:31 PM
T         |SYS_P123420   |                 2|         1|11-APR-2011 12:21:37 PM
T         |SYS_P123421   |                 3|         1|11-APR-2011 12:21:37 PM
T         |SYS_P123422   |                 4|         1|11-APR-2011 12:21:37 PM
T         |SYS_P123423   |                 5|         1|11-APR-2011 12:21:37 PM

SQL>
SQL> SELECT index_name, num_rows, last_analyzed
  2    FROM user_indexes
  3   WHERE table_name = 'T';

INDEX_NAME|  NUM_ROWS|LAST_ANALYZED
----------|----------|-----------------------
PK_T      |         4|11-APR-2011 12:21:38 PM

SQL>
SQL> EXEC dbms_lock.sleep(2);

PL/SQL procedure successfully completed.

SQL>
SQL> prompt ==>No insert here, just gather stats
==>No insert here, just gather stats
SQL>
SQL> BEGIN
  2    DBMS_STATS.gather_table_stats (ownname	       => USER,
  3  				      tabname	       => 'T',
  4  				      estimate_percent => DBMS_STATS.auto_sample_size,
  5  				      method_opt       => 'FOR ALL COLUMNS SIZE 1',
  6  				      granularity      => 'AUTO',
  7  				      degree	       => 2,
  8  				      cascade	       => TRUE,
  9  				      no_invalidate    => TRUE);
 10  END;
 11  /

PL/SQL procedure successfully completed.

SQL>
SQL> SELECT table_name, num_rows, last_analyzed
  2    FROM user_tables
  3   WHERE table_name = 'T';

TABLE_NAME|  NUM_ROWS|LAST_ANALYZED
----------|----------|-----------------------
T         |         4|11-APR-2011 12:21:41 PM

SQL>
SQL>   SELECT table_name,
  2  	      partition_name,
  3  	      partition_position,
  4  	      num_rows,
  5  	      last_analyzed
  6  	 FROM user_tab_partitions
  7  	WHERE table_name = 'T'
  8  ORDER BY partition_position;

TABLE_NAME|PARTITION_NAME|PARTITION_POSITION|  NUM_ROWS|LAST_ANALYZED
----------|--------------|------------------|----------|-----------------------
T         |P1            |                 1|         0|11-APR-2011 12:21:31 PM
T         |SYS_P123420   |                 2|         1|11-APR-2011 12:21:41 PM
T         |SYS_P123421   |                 3|         1|11-APR-2011 12:21:41 PM
T         |SYS_P123422   |                 4|         1|11-APR-2011 12:21:41 PM
T         |SYS_P123423   |                 5|         1|11-APR-2011 12:21:41 PM

SQL>
SQL> SELECT index_name, num_rows, last_analyzed
  2    FROM user_indexes
  3   WHERE table_name = 'T';

INDEX_NAME|  NUM_ROWS|LAST_ANALYZED
----------|----------|-----------------------
PK_T      |         4|11-APR-2011 12:21:41 PM

SQL>
SQL> exit
]]> By: Greg Rahn http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-17228 Greg Rahn Wed, 06 Apr 2011 17:57:21 +0000 http://structureddata.org/?p=70#comment-17228 @DBA Can you provide a test case that demonstrates this behavior? @DBA

Can you provide a test case that demonstrates this behavior?

]]> By: DBA http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-17210 DBA Tue, 05 Apr 2011 15:40:23 +0000 http://structureddata.org/?p=70#comment-17210 I tested the incremental stats feature with 11.1.0.7 version on Linux 64 bit OS. Sometimes it works, sometimes it doesn't. And the only thing I have been able to find so far is that if you have a Primary key (with a global index) on a partitioned table that is not part of the partition key (e.g. in OLTP), the incremental stats don't work. I have gathered stats on a table back to back that was set to use incremental stats. It gathered stats on all the partitions all over again the second time even when there was not a single DML that happened in between the two stat gathering. I have seen a few bugs about this on Mysupport website. Have you anything to share about this phenomenon? I tested the incremental stats feature with 11.1.0.7 version on Linux 64 bit OS. Sometimes it works, sometimes it doesn’t. And the only thing I have been able to find so far is that if you have a Primary key (with a global index) on a partitioned table that is not part of the partition key (e.g. in OLTP), the incremental stats don’t work. I have gathered stats on a table back to back that was set to use incremental stats. It gathered stats on all the partitions all over again the second time even when there was not a single DML that happened in between the two stat gathering. I have seen a few bugs about this on Mysupport website. Have you anything to share about this phenomenon?

]]> By: Data Warehousing and Statistics in Oracle 11g – incremental global statistics « RNM http://structureddata.org/2008/07/16/oracle-11g-incremental-global-statistics-on-partitioned-tables/#comment-15314 Data Warehousing and Statistics in Oracle 11g – incremental global statistics « RNM Thu, 30 Dec 2010 23:40:02 +0000 http://structureddata.org/?p=70#comment-15314 [...] Greg Rahn – Oracle 11g: Incremental Global Statistics On Partitioned Tables [...] [...] Greg Rahn – Oracle 11g: Incremental Global
Statistics On Partitioned Tables [...]

]]>