MySQL中關於not in和minus應用的優化。本站提示廣大學習愛好者：（MySQL中關於not in和minus應用的優化）文章只能為提供參考，不一定能成為您想要的結果。以下是MySQL中關於not in和minus應用的優化正文
優化前：
select count(t.id)
from test t
where t.status = 1
and t.id not in (select distinct a.app_id
from test2 a
where a.type = 1
and a.rule_id in (152, 153, 154))
17:20:57 laojiu>@plan
PLAN_TABLE_OUTPUT
————————————————————————————————————————-
Plan hash value: 684502086
—————————————————————————————-
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
—————————————————————————————-
| 0 | SELECT STATEMENT | | 1 | 18 | 176K (2)| 00:35:23 |
| 1 | SORT AGGREGATE | | 1 | 18 | | |
|* 2 | FILTER | | | | | |
|* 3 | TABLE ACCESS FULL| test | 1141 | 20538 | 845 (2)| 00:00:11 |
|* 4 | TABLE ACCESS FULL| test2 | 1 | 12 | 309 (2)| 00:00:04 |
—————————————————————————————-
Predicate Information (identified by operation id):
—————————————————
2 – filter( NOT EXISTS (SELECT /*+ */ 0 FROM “test2″ “A” WHERE
“A”.”type”=1 AND (“A”.”RULE_ID”=152 OR “A”.”RULE_ID”=153 OR
“A”.”RULE_ID”=154) AND LNNVL(“A”.”APP_ID”<>:B1)))
3 – filter(“T”.”status”=1)
4 – filter(“A”.”type”=1 AND (“A”.”RULE_ID”=152 OR “A”.”RULE_ID”=153 OR
“A”.”RULE_ID”=154) AND LNNVL(“A”.”APP_ID”<>:B1))
Statistics
———————————————————-
0 recursive calls
0 db block gets
1762169 consistent gets
0 physical reads
0 redo size
519 bytes sent via SQL*Net to client
492 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
1 rows processed
21 rows selected.
優化後：
select count(*) from(
select t.id
from test t
where t.status = 1
minus
select distinct a.app_id
from test2 a
where a.type = 1
and a.rule_id in (152, 153, 154))
17:23:33 laojiu>@plan
PLAN_TABLE_OUTPUT
————————————————————————————————————————-
Plan hash value: 631655686
————————————————————————————————–
| Id | Operation | Name | Rows | Bytes |TempSpc| Cost (%CPU)| Time |
————————————————————————————————–
| 0 | SELECT STATEMENT | | 1 | | | 1501 (2)| 00:00:19 |
| 1 | SORT AGGREGATE | | 1 | | | | |
| 2 | VIEW | | 1141 | | | 1501 (2)| 00:00:19 |
| 3 | MINUS | | | | | | |
| 4 | SORT UNIQUE | | 1141 | 20538 | | 846 (2)| 00:00:11 |
|* 5 | TABLE ACCESS FULL| test | 1141 | 20538 | | 845 (2)| 00:00:11 |
| 6 | SORT UNIQUE | | 69527 | 814K| 3632K| 654 (2)| 00:00:08 |
|* 7 | TABLE ACCESS FULL| test2 | 84140 | 986K| | 308 (2)| 00:00:04 |
————————————————————————————————–
Predicate Information (identified by operation id):
—————————————————
5 – filter(“T”.”status”=1)
7 – filter(“A”.”type”=1 AND (“A”.”RULE_ID”=152 OR “A”.”RULE_ID”=153 OR
“A”.”RULE_ID”=154))
21 rows selected.
Statistics
———————————————————-
1 recursive calls
0 db block gets
2240 consistent gets
0 physical reads
0 redo size
516 bytes sent via SQL*Net to client
492 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
2 sorts (memory)
0 sorts (disk)
1 rows processed
在優化sql的時刻，我們須要改變一下思緒，等價的改寫sql；
改寫後的sql因為邏輯讀獲得了天崩地裂翻天覆地的轉變，很快獲得成果。
第一條sql履行籌劃中有一個函數，LNNVL(“A”.”APP_ID”<>:B1)，lnnvl(exp)
假如exp的成果是false或許是unknown,那末lnnvl前往true;
假如exp的成果是true,前往false.
