Comments on: GATHER_SCHEMA_STATS & ORA-03001: unimplemented feature http://amardeepsidhu.com/blog/2009/07/08/gather_schema_stats-ora-03001-unimplemented-feature/ My experiences with technology... Sat, 20 Mar 2010 03:30:46 +0000 http://wordpress.org/?v=2.9.1 hourly 1 By: Blogroll Report 03/07/2009 – 10/07/2006 « Coskan’s Approach to Oracle http://amardeepsidhu.com/blog/2009/07/08/gather_schema_stats-ora-03001-unimplemented-feature/comment-page-1/#comment-7845 Blogroll Report 03/07/2009 – 10/07/2006 « Coskan’s Approach to Oracle Fri, 10 Jul 2009 18:42:34 +0000 http://amardeepsidhu.com/blog/?p=233#comment-7845 [...] Amardeep Sidhu – GATHER_SCHEMA_STATS & ORA-03001: unimplemented feature [...] [...] Amardeep Sidhu – GATHER_SCHEMA_STATS & ORA-03001: unimplemented feature [...]

]]> By: Sidhu http://amardeepsidhu.com/blog/2009/07/08/gather_schema_stats-ora-03001-unimplemented-feature/comment-page-1/#comment-7831 Sidhu Thu, 09 Jul 2009 16:59:42 +0000 http://amardeepsidhu.com/blog/?p=233#comment-7831 Thanks man ! We never cross each other in this area of topics for technical blogs ;) Thanks man ! We never cross each other in this area of topics for technical blogs ;)

]]> By: Vaibhav Garg http://amardeepsidhu.com/blog/2009/07/08/gather_schema_stats-ora-03001-unimplemented-feature/comment-page-1/#comment-7830 Vaibhav Garg Thu, 09 Jul 2009 16:04:14 +0000 http://amardeepsidhu.com/blog/?p=233#comment-7830 Wow very neat and clean post. Didn't understand much but it looks great :) Wow very neat and clean post.

Didn’t understand much but it looks great :)

]]> By: Sidhu http://amardeepsidhu.com/blog/2009/07/08/gather_schema_stats-ora-03001-unimplemented-feature/comment-page-1/#comment-7828 Sidhu Thu, 09 Jul 2009 15:23:43 +0000 http://amardeepsidhu.com/blog/?p=233#comment-7828 @ Sarayu & Coskan I didn't create this index ;). But to best of my knowledge it has been created to index NULL entries. Something like this: <pre>SQL> create table st1 as select * from all_objects; Table created. SQL> select count(*),status 2 from st1 3 group by status; COUNT(*) STATUS ---------- ------- 1 INVALID 23559 VALID SQL> update st1 set status=null where rownum < 15; 14 rows updated. SQL> commit; Commit complete. SQL> create index ind1_st1 on st1(status); Index created. SQL> exec dbms_stats.gather_table_stats(user,'ST1',cascade=>true); PL/SQL procedure successfully completed. SQL> explain plan for select * from st1 where status is null; Explained. SQL> @e PLAN_TABLE_OUTPUT ------------------------------------------------------------------------------------ -------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost | -------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 14 | 1302 | 33 | |* 1 | TABLE ACCESS FULL | ST1 | 14 | 1302 | 33 | -------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 1 - filter("ST1"."STATUS" IS NULL) Note: cpu costing is off 14 rows selected. SQL> create index ind1_st2 on st1(status,1); Index created. SQL> explain plan for select * from st1 where status is null; Explained. SQL> @e PLAN_TABLE_OUTPUT ------------------------------------------------------------------------------------ --------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost | --------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 14 | 1302 | 3 | | 1 | TABLE ACCESS BY INDEX ROWID| ST1 | 14 | 1302 | 3 | |* 2 | INDEX RANGE SCAN | IND1_ST2 | 14 | | 2 | --------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 2 - access("ST1"."STATUS" IS NULL) Note: cpu costing is off 15 rows selected. SQL> drop index ind1_st2; Index dropped. SQL> create index ind1_st2 on st1(status,'1'); Index created. SQL> explain plan for select * from st1 where status is null; Explained. SQL> @e PLAN_TABLE_OUTPUT ------------------------------------------------------------------------------------ --------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost | --------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 14 | 1302 | 3 | | 1 | TABLE ACCESS BY INDEX ROWID| ST1 | 14 | 1302 | 3 | |* 2 | INDEX RANGE SCAN | IND1_ST2 | 14 | | 2 | --------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 2 - access("ST1"."STATUS" IS NULL) Note: cpu costing is off 15 rows selected. SQL></pre> @ Sarayu & Coskan

I didn’t create this index ;) . But to best of my knowledge it has been created to index NULL entries. Something like this:

SQL> create table st1 as select * from all_objects;

Table created.

SQL> select count(*),status
  2  from st1
  3  group by status;

  COUNT(*) STATUS
---------- -------
         1 INVALID
     23559 VALID

SQL> update st1 set status=null where rownum < 15;

14 rows updated.

SQL> commit;

Commit complete.

SQL> create index ind1_st1 on st1(status);

Index created.

SQL> exec dbms_stats.gather_table_stats(user,'ST1',cascade=>true);

PL/SQL procedure successfully completed.

SQL> explain plan for select * from st1 where status is null;

Explained.

SQL> @e

PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------

--------------------------------------------------------------------
| Id  | Operation            |  Name       | Rows  | Bytes | Cost  |
--------------------------------------------------------------------
|   0 | SELECT STATEMENT     |             |    14 |  1302 |    33 |
|*  1 |  TABLE ACCESS FULL   | ST1         |    14 |  1302 |    33 |
--------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   1 - filter("ST1"."STATUS" IS NULL)

Note: cpu costing is off

14 rows selected.

SQL> create index ind1_st2 on st1(status,1);

Index created.

SQL> explain plan for select * from st1 where status is null;

Explained.

SQL> @e

PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------

---------------------------------------------------------------------------
| Id  | Operation                   |  Name       | Rows  | Bytes | Cost  |
---------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |             |    14 |  1302 |     3 |
|   1 |  TABLE ACCESS BY INDEX ROWID| ST1         |    14 |  1302 |     3 |
|*  2 |   INDEX RANGE SCAN          | IND1_ST2    |    14 |       |     2 |
---------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - access("ST1"."STATUS" IS NULL)

Note: cpu costing is off

15 rows selected.

SQL> drop index ind1_st2;

Index dropped.

SQL> create index ind1_st2 on st1(status,'1');

Index created.

SQL> explain plan for select * from st1 where status is null;

Explained.

SQL> @e

PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------

---------------------------------------------------------------------------
| Id  | Operation                   |  Name       | Rows  | Bytes | Cost  |
---------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |             |    14 |  1302 |     3 |
|   1 |  TABLE ACCESS BY INDEX ROWID| ST1         |    14 |  1302 |     3 |
|*  2 |   INDEX RANGE SCAN          | IND1_ST2    |    14 |       |     2 |
---------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - access("ST1"."STATUS" IS NULL)

Note: cpu costing is off

15 rows selected.

SQL>
]]> By: Coskan http://amardeepsidhu.com/blog/2009/07/08/gather_schema_stats-ora-03001-unimplemented-feature/comment-page-1/#comment-7816 Coskan Thu, 09 Jul 2009 01:12:00 +0000 http://amardeepsidhu.com/blog/?p=233#comment-7816 My question is what is this function based index actually doing? My question is what is this function based index actually doing?

]]> By: sarayu http://amardeepsidhu.com/blog/2009/07/08/gather_schema_stats-ora-03001-unimplemented-feature/comment-page-1/#comment-7812 sarayu Wed, 08 Jul 2009 20:59:50 +0000 http://amardeepsidhu.com/blog/?p=233#comment-7812 Why do you ever want to create index as test1(comm,1); Any special reasons?? Why do you ever want to create index as test1(comm,1);
Any special reasons??

]]>