MySQL中对于not in和minus使用的优化

优化前：

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.

优化后：

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

在优化sql的时候，我们需要转变一下思路，等价的改写sql；

改写后的sql由于逻辑读得到了天翻地覆的改变，很快得到结果。

第一条sql执行计划中有一个函数，LNNVL(“A”.”APP_ID”<>:B1)，lnnvl(exp)

如果exp的结果是false或者是unknown,那么lnnvl返回true;

如果exp的结果是true,返回false.