Printable View
When I do a count of all the rows in a table we have called production.dbug it tell me 1051
When I select all_tables.num_rows where table name = production.dbug it returns 937807
Why the huge difference?
It seems as if all_tables may be storing a high water mark?
Is this bad? The table is cleared every now and then( delete * ) should I do a truncate at that time?
Thanks,
SM
Truncate is a good way to reset the HWM. This is only possible if you are deleting all rows. Remember truncate is a DDL and cannot be rolled back.
Nirasha
is that what num_rows in all_tables is, a high water mark?
Thanks again,
SM
*bump*
Count * actually counts the indexes whereas select all will count each row on the table.
DBA_TABLES reflects the row counts as of the last time the table was analyzed. Issuing a select count(*) from tabxyz will give you the exact number of rows in the table. For example:
SQL> create table xyz (x char(10), y char(10), z char(10));
Table created.
SQL> insert into xyz values ('x','y','z');
1 row created.
SQL> insert into xyz values ('x','y','z');
1 row created.
SQL> insert into xyz values ('x','y','z');
1 row created.
SQL> insert into xyz values ('x','y','z');
1 row created.
SQL> analyze table xyz compute statistics;
Table analyzed.
SQL> select table_name, num_rows from user_tables;
TABLE_NAME NUM_ROWS
------------------------------ ----------
XYZ 4
SQL> insert into xyz values ('x','y','z');
1 row created.
SQL> insert into xyz values ('x','y','z');
1 row created.
SQL> insert into xyz values ('x','y','z');
1 row created.
SQL> insert into xyz values ('x','y','z');
1 row created.
SQL> commit;
Commit complete.
SQL> select count(*) from xyz;
COUNT(*)
----------
8
SQL> select table_name, num_rows from user_tables;
TABLE_NAME NUM_ROWS
------------------------------ ----------
XYZ 4
SQL> analyze table xyz compute statistics;
Table analyzed.
SQL> select table_name, num_rows from user_tables;
TABLE_NAME NUM_ROWS
------------------------------ ----------
XYZ 8
