Database stats

Question

Hi,
I want to know if I disable stats gathering in the database then what would the optimizer do when a query will be executed? will it calculate statistics on run time? If yes then what will be the overhead?. Also suggest me for large table is it recommended to calculate stats as it take long time. If some can provide a brief explanation on this then I will be very thankful, i tried checking to Google but cannot find relevant material.
Thanks in advance.

Answer 1

9135487 wrote:
Thanks Aman, I have Oracle 11.2.0.3.0
Stat gathering is not stopped but I just wanted to know what if one has to stop and and what would be the impact of stopping stats on query execution. So now if i stop it no dynamic sampling would be done? as the CBO will use old stats? What would you suggest in this case?Have a look and hope it clears your doubt,
Connected to:
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
With the Partitioning, Automatic Storage Management, OLAP, Data Mining
and Real Application Testing options
SQL> create table t as select * from scott.emp;
Table created.
SQL> set autot trace exp
SQL> select * from t;
Execution Plan
Plan hash value: 1601196873
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT  |      |    14 |  1218 |     2   (0)| 00:00:01 |
|   1 |  TABLE ACCESS FULL| T    |    14 |  1218 |     2   (0)| 00:00:01 |
Note
   - dynamic sampling used for this statement (level=2)
SQL> exec dbms_stats.gather_table_stats(user,'T');
PL/SQL procedure successfully completed.
SQL> select * from t;
Execution Plan
Plan hash value: 1601196873
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT  |      |    14 |   532 |     2   (0)| 00:00:01 |
|   1 |  TABLE ACCESS FULL| T    |    14 |   532 |     2   (0)| 00:00:01 |
SQL> insert into t select * from t;
14 rows created.
Execution Plan
Plan hash value: 1601196873
| Id  | Operation                | Name | Rows  | Bytes | Cost (%CPU)| Time
|
|   0 | INSERT STATEMENT         |      |    14 |   532 |     2   (0)| 00:00:01
|
|   1 |  LOAD TABLE CONVENTIONAL | T    |       |       |            |
|
|   2 |   TABLE ACCESS FULL      | T    |    14 |   532 |     2   (0)| 00:00:01
|
SQL> /
28 rows created.
Execution Plan
Plan hash value: 1601196873
| Id  | Operation                | Name | Rows  | Bytes | Cost (%CPU)| Time
|
|   0 | INSERT STATEMENT         |      |    14 |   532 |     2   (0)| 00:00:01
|
|   1 |  LOAD TABLE CONVENTIONAL | T    |       |       |            |
|
|   2 |   TABLE ACCESS FULL      | T    |    14 |   532 |     2   (0)| 00:00:01
|
SQL> commit; 
Commit complete.
SQL> set autot off
SQL> select num_rows from user_tables where table_name='T';
  NUM_ROWS
        14
SQL> select count(*) from t;
  COUNT(*)
        56
SQL> set autot trace exp
SQL> select * from t;
Execution Plan
Plan hash value: 1601196873
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT  |      |    14 |   532 |     2   (0)| 00:00:01 |
|   1 |  TABLE ACCESS FULL| T    |    14 |   532 |     2   (0)| 00:00:01 |
SQL> So as you should be able to see, the CBO still believes that I have got just 14 rows in my table whereas the actual count is far higher. So obviously I would be required to refresh the stats again.
HTH
Aman....