由 dawei
网络上很多谣言是后面跟小表用not in，后面跟大表用not exists,难道真的是这样子的嘛？
情况下面测试：
1、先用小表测试（1000条记录和一张8万条记录的表）：
SQL> create table not_in_s as select * from dba_objects where object_id <= 10000;
Table created.
Elapsed: 00:00:00.17
SQL> create table not_exists_s as select * from dba_objects where object_id <= 10000;
Table created.
查看执行计划：
小表not in
set autotrace traceonly explain
SQL> select * from dba_objects where object_id
not in (select object_id from not_exists_s);Elapsed: 00:00:00.06
Execution Plan----------------------------------------------------------Plan hash value: 566308792
-----------------------------------------------------------------------------------------------| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |-----------------------------------------------------------------------------------------------| 0 | SELECT STATEMENT | | 87028 | 18M| 337 (1)| 00:00:05 ||* 1 | HASH JOIN RIGHT ANTI NA | | 87028 | 18M| 337 (1)| 00:00:05 || 2 | TABLE ACCESS FULL | NOT_EXISTS_S | 11495 | 145K| 36 (0)| 00:00:01 || 3 | VIEW | DBA_OBJECTS | 87036 | 17M| 301 (1)| 00:00:04 || 4 | UNION-ALL | | | | | ||* 5 | TABLE ACCESS BY INDEX ROWID| SUM$ | 1 | 9 | 1 (0)| 00:00:01 ||* 6 | INDEX UNIQUE SCAN | I_SUM$_1 | 1 | | 0 (0)| 00:00:01 || 7 | TABLE ACCESS BY INDEX ROWID| OBJ$ | 1 | 30 | 3 (0)| 00:00:01 ||* 8 | INDEX RANGE SCAN | I_OBJ1 | 1 | | 2 (0)| 00:00:01 ||* 9 | FILTER | | | | | ||* 10 | HASH JOIN | | 87035 | 10M| 299 (2)| 00:00:04 || 11 | TABLE ACCESS CLUSTER | USER$ | 92 | 1564 | 3 (0)| 00:00:01 || 12 | INDEX FULL SCAN | I_USER# | 1 | | 1 (0)| 00:00:01 ||* 13 | HASH JOIN | | 87035 | 9009K| 296 (2)| 00:00:04 || 14 | INDEX FULL SCAN | I_USER2 | 92 | 2024 | 1 (0)| 00:00:01 ||* 15 | TABLE ACCESS FULL | OBJ$ | 87035 | 7139K| 294 (1)| 00:00:04 || 16 | NESTED LOOPS | | 1 | 29 | 2 (0)| 00:00:01 ||* 17 | INDEX SKIP SCAN | I_USER2 | 1 | 20 | 1 (0)| 00:00:01 ||* 18 | INDEX RANGE SCAN | I_OBJ4 | 1 | 9 | 1 (0)| 00:00:01 || 19 | NESTED LOOPS | | 1 | 105 | 2 (0)| 00:00:01 || 20 | TABLE ACCESS FULL | LINK$ | 1 | 88 | 2 (0)| 00:00:01 || 21 | TABLE ACCESS CLUSTER | USER$ | 1 | 17 | 0 (0)| 00:00:01 ||* 22 | INDEX UNIQUE SCAN | I_USER# | 1 | | 0 (0)| 00:00:01 |-----------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):---------------------------------------------------
1 - access("OBJECT_ID"="OBJECT_ID") 5 - filter(BITAND("S"."XPFLAGS",8388608)=8388608) 6 - access("S"."OBJ#"=:B1) 8 - access("EO"."OBJ#"=:B1) 9 - filter("O"."TYPE#"<>4 AND "O"."TYPE#"<>5 AND "O"."TYPE#"<>7 AND "O"."TYPE#"<>8 AND "O"."TYPE#"<>9 AND "O"."TYPE#"<>11 AND "O"."TYPE#"<>12 AND "O"."TYPE#"<>13 AND "O"."TYPE#"<>14 AND "O"."TYPE#"<>22 AND "O"."TYPE#"<>87 AND "O"."TYPE#"<>88 OR BITAND("U"."SPARE1",16)=0 OR ("O"."TYPE#"=4 OR "O"."TYPE#"=5 OR "O"."TYPE#"=7 OR "O"."TYPE#"=8 OR "O"."TYPE#"=9 OR "O"."TYPE#"=10 OR "O"."TYPE#"=11 OR "O"."TYPE#"=12 OR "O"."TYPE#"=13 OR "O"."TYPE#"=14 OR "O"."TYPE#"=22 OR "O"."TYPE#"=87) AND (SYS_CONTEXT('userenv','current_edition_name')='ORA$BASE' AND "U"."TYPE#"<>2 OR "U"."TYPE#"=2 AND "U"."SPARE2"=TO_NUMBER(SYS_CONTEXT('userenv','current_edition_id')) OR EXISTS (SELECT 0 FROM SYS."USER$" "U2",SYS."OBJ$" "O2" WHERE "O2"."OWNER#"="U2"."USER#" AND "O2"."TYPE#"=88 AND "O2"."DATAOBJ#"=:B1 AND "U2"."TYPE#"=2 AND "U2"."SPARE2"=TO_NUMBER(SYS_CONTEXT('userenv','current_edition_id')) ))) 10 - access("O"."SPARE3"="U"."USER#") 13 - access("O"."OWNER#"="U"."USER#") 15 - filter("O"."TYPE#"<>10 AND "O"."NAME"<>'_NEXT_OBJECT' AND "O"."NAME"<>'_default_auditing_options_' AND "O"."LINKNAME" IS NULL AND BITAND("O"."FLAGS",128)=0) 17 - access("U2"."TYPE#"=2 AND "U2"."SPARE2"=TO_NUMBER(SYS_CONTEXT('userenv','current _edition_id'))) filter("U2"."TYPE#"=2 AND "U2"."SPARE2"=TO_NUMBER(SYS_CONTEXT('userenv','current _edition_id'))) 18 - access("O2"."DATAOBJ#"=:B1 AND "O2"."TYPE#"=88 AND "O2"."OWNER#"="U2"."USER#") 22 - access("L"."OWNER#"="U"."USER#")
Note----- - dynamic sampling used for this statement (level=2)
小表 not exists:
SQL> select * from dba_objects where not exists (select object_id from not_in_s where not_in_s.object_id = dba_objects.object_id);Elapsed: 00:00:00.04
Execution Plan----------------------------------------------------------Plan hash value: 4002778081
----------------------------------------------------------------------------------------------| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |----------------------------------------------------------------------------------------------| 0 | SELECT STATEMENT | | 87027 | 18M| 337 (1)| 00:00:05 ||* 1 | HASH JOIN RIGHT ANTI | | 87027 | 18M| 337 (1)| 00:00:05 || 2 | TABLE ACCESS FULL | NOT_IN_S | 9749 | 123K| 36 (0)| 00:00:01 || 3 | VIEW | DBA_OBJECTS | 87036 | 17M| 301 (1)| 00:00:04 || 4 | UNION-ALL | | | | | ||* 5 | TABLE ACCESS BY INDEX ROWID| SUM$ | 1 | 9 | 1 (0)| 00:00:01 ||* 6 | INDEX UNIQUE SCAN | I_SUM$_1 | 1 | | 0 (0)| 00:00:01 || 7 | TABLE ACCESS BY INDEX ROWID| OBJ$ | 1 | 30 | 3 (0)| 00:00:01 ||* 8 | INDEX RANGE SCAN | I_OBJ1 | 1 | | 2 (0)| 00:00:01 ||* 9 | FILTER | | | | | ||* 10 | HASH JOIN | | 87035 | 10M| 299 (2)| 00:00:04 || 11 | TABLE ACCESS CLUSTER | USER$ | 92 | 1564 | 3 (0)| 00:00:01 || 12 | INDEX FULL SCAN | I_USER# | 1 | | 1 (0)| 00:00:01 ||* 13 | HASH JOIN | | 87035 | 9009K| 296 (2)| 00:00:04 || 14 | INDEX FULL SCAN | I_USER2 | 92 | 2024 | 1 (0)| 00:00:01 ||* 15 | TABLE ACCESS FULL | OBJ$ | 87035 | 7139K| 294 (1)| 00:00:04 || 16 | NESTED LOOPS | | 1 | 29 | 2 (0)| 00:00:01 ||* 17 | INDEX SKIP SCAN | I_USER2 | 1 | 20 | 1 (0)| 00:00:01 ||* 18 | INDEX RANGE SCAN | I_OBJ4 | 1 | 9 | 1 (0)| 00:00:01 || 19 | NESTED LOOPS | | 1 | 105 | 2 (0)| 00:00:01 || 20 | TABLE ACCESS FULL | LINK$ | 1 | 88 | 2 (0)| 00:00:01 || 21 | TABLE ACCESS CLUSTER | USER$ | 1 | 17 | 0 (0)| 00:00:01 ||* 22 | INDEX UNIQUE SCAN | I_USER# | 1 | | 0 (0)| 00:00:01 |----------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):---------------------------------------------------
1 - access("NOT_IN_S"."OBJECT_ID"="DBA_OBJECTS"."OBJECT_ID") 5 - filter(BITAND("S"."XPFLAGS",'current_edition_id') )))) 10 - access("O"."SPARE3"="U"."USER#") 13 - access("O"."OWNER#"="U"."USER#") 15 - filter("O"."TYPE#"<>10 AND "O"."NAME"<>'_NEXT_OBJECT' AND "O"."NAME"<>'_default_auditing_options_' AND "O"."LINKNAME" IS NULL AND BITAND("O"."FLAGS",'curren t_edition_id'))) filter("U2"."TYPE#"=2 AND "U2"."SPARE2"=TO_NUMBER(SYS_CONTEXT('userenv','curren t_edition_id'))) 18 - access("O2"."DATAOBJ#"=:B1 AND "O2"."TYPE#"=88 AND "O2"."OWNER#"="U2"."USER#") 22 - access("L"."OWNER#"="U"."USER#")
Note----- - dynamic sampling used for this statement (level=2)
1000行记录的小表用not in，not exists 消耗的成本都是337。
2、先用大表测试（5万条记录和一张8万条记录的表）：
SQL> create table not_in_b as select * from dba_objects where object_id <= 50000;create table not_exists_b as select * from dba_objects where object_id <= 50000;Table created.
Elapsed: 00:00:00.55SQL>
Table created.
Elapsed: 00:00:00.50
大表not in:
SQL> select * from dba_objects where object_id not in (select object_id from not_exists_b);Elapsed: 00:00:00.07
Execution Plan----------------------------------------------------------Plan hash value: 3301458141