一种高效率的update多表关联更新测试

SQL> set echo on SQL> set time on 17:56:09 SQL> set timing on 17:56:12 SQL> set autotrace traceonly 17:56:19 SQL> @aaa.sql 17:56:21 SQL> update ( 17:56:21   2    select /*+use_hash(a,b)*/a.bill_bal bill_bal_a ,b.bill_bal bill_bal_b 17:56:21   3    from tjw_target_user a,tmp_tjw b 17:56:21   4    where a.bill_id = b.bill_id 17:56:21   5  ) set bill_bal_a = bill_bal_b; 1335068 rows updated. Elapsed: 00:02:52.74 Execution Plan ----------------------------------------------------------    0      UPDATE STATEMENT Optimizer=ALL_ROWS (Cost=681 Card=82 Bytes=           2952)    1    0   UPDATE OF 'TJW_TARGET_USER'    2    1     HASH JOIN (Cost=681 Card=82 Bytes=2952)    3    2       TABLE ACCESS (FULL) OF 'TMP_TJW' (Cost=2 Card=82 Bytes           =1804)    4    2       TABLE ACCESS (FULL) OF 'TJW_TARGET_USER' (Cost=674 Car           d=1540412 Bytes=21565768) Statistics ----------------------------------------------------------         610  recursive calls     1365395  db block gets       47221  consistent gets       18052  physical reads   318042748  redo size         495  bytes sent via SQL*Net to client         671  bytes received via SQL*Net from client           3  SQL*Net roundtrips to/from client           1  sorts (memory)           0  sorts (disk)     1335068  rows processed 17:59:14 SQL> 17:59:14 SQL> commit; Commit complete. Elapsed: 00:00:00.04 17:59:14 SQL> 17:59:14 SQL> update tjw_target_user a  set bill_bal = ( 17:59:14   2          select bill_bal from tmp_tjw b where a.bill_id = b.bill_id 17:59:14   3  ) 17:59:14   4  where exists ( 17:59:14   5     select 1 from tmp_tjw c where a.bill_id = c.bill_id 17:59:14   6  ); 1335068 rows updated. Elapsed: 00:07:08.56 Execution Plan ----------------------------------------------------------    0      UPDATE STATEMENT Optimizer=ALL_ROWS (Cost=169 Card=82 Bytes=           1886)    1    0   UPDATE OF 'TJW_TARGET_USER'    2    1     NESTED LOOPS (Cost=169 Card=82 Bytes=1886)    3    2       SORT (UNIQUE)    4    3         INDEX (FAST FULL SCAN) OF 'PK_TMP_TJW_IDX' (UNIQUE)           (Cost=2 Card=82 Bytes=738)    5    2       INDEX (RANGE SCAN) OF 'PK_TJW_TARGET_USER_IDX' (UNIQUE           ) (Cost=2 Card=1 Bytes=14)    6    1     TABLE ACCESS (BY INDEX ROWID) OF 'TMP_TJW' (Cost=1 Card=           1 Bytes=22)    7    6       INDEX (RANGE SCAN) OF 'PK_TMP_TJW_IDX' (UNIQUE) (Cost=           1 Card=1) Statistics ----------------------------------------------------------         140  recursive calls     1364964  db block gets     6713317  consistent gets       19867  physical reads   317737072  redo size         506  bytes sent via SQL*Net to client         672  bytes received via SQL*Net from client           3  SQL*Net roundtrips to/from client           1  sorts (memory)           1  sorts (disk)     1335068  rows processed 注意 : 需要在a.bill_id, b.bill_id列上有唯一约束或索引才可以,否则报错: 实际例子: update cm_cb_ss_circuitendpoint_bak c set c.sys_int_id=(select max(a.sys_int_id) from (select * from cm_cb_ss_circuitendpoint_bak where UPPER(data_status)!='NEW' and data_status is not null) a where a.dn=c.dn) where c.dn in (select dn from cm_cb_ss_circuitendpoint_bak where UPPER(data_status)='NEW' or data_status is null) 这样就可以了。

你可能感兴趣的:(sql,user,table,Access,statistics,loops)