两个左连接SQL执行计划解析（Oracle和PGSQL对比）：

上一篇解析链接如下：

https://www.cnblogs.com/wcwen1990/p/9325968.html

1、SQL示例1：

SQL> select *

from (

select * from tmp1 where c >= 1

) t1 left join (

select * from tmp2 where b < 30

) t2 on t1.a = t2.a

and t2.d > 1 and t1.e >= 2

where t1.b < 50

;

A        B           C      E         A        B       D          E

---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------
      2       20           2      2         2           20       2          2
      4       40           4      4
      3       30           3      3
      1       10           1      1

Execution Plan

----------------------------------------------------------

Plan hash value: 2592321047

---------------------------------------------------------------------------

| Id  | Operation       | Name | Rows  | Bytes | Cost (%CPU)| Time      |

---------------------------------------------------------------------------

|   0 | SELECT STATEMENT   |      |    4 |   416 |    7  (15)| 00:00:01 |

|*  1 |  HASH JOIN OUTER   |      |    4 |   416 |    7  (15)| 00:00:01 |

|*  2 |   TABLE ACCESS FULL| TMP1 |    4 |   208 |    3   (0)| 00:00:01 |

|*  3 |   TABLE ACCESS FULL| TMP2 |    1 |    52 |    3   (0)| 00:00:01 |

---------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

1 - access("TMP1"."A"="TMP2"."A"(+))
        filter("TMP1"."E">=CASE    WHEN ("TMP2"."A"(+) IS NOT NULL) THEN 2
           ELSE 2 END )
    2 - filter("TMP1"."B"<50 AND "C">=1)
    3 - filter("TMP2"."D"(+)>1 AND "B"(+)<30)

Note

-----
    - dynamic sampling used for this statement (level=2)

Statistics

----------------------------------------------------------
       0  recursive calls
       0  db block gets
       7  consistent gets
       0  physical reads
       0  redo size
        1082  bytes sent via SQL*Net to client
     524  bytes received via SQL*Net from client
       2  SQL*Net roundtrips to/from client
       0  sorts (memory)
       0  sorts (disk)
       4  rows processed

postgres=# explain analyze select *

postgres-# from (

postgres(# select * from tmp1 where c >= 1

postgres(# ) t1 left join (

postgres(# select * from tmp2 where b < 30

postgres(# ) t2 on t1.a = t2.a

postgres-# and t2.d > 1 and t1.e >= 2

postgres-# where t1.b < 50

postgres-# ;
                                                   QUERY PLAN                                                 

--------------------------------------------------------------------------------------------------------------
  Hash Left Join  (cost=34.90..80.00 rows=181 width=32) (actual time=0.021..0.035 rows=4 loops=1)
    Hash Cond: ("outer".a = "inner".a)
    Join Filter: ("outer".e >= 2)
    ->  Seq Scan on tmp1  (cost=0.00..34.45 rows=181 width=16) (actual time=0.006..0.011 rows=4 loops=1)
          Filter: ((c >= 1) AND (b < 50))
    ->  Hash  (cost=34.45..34.45 rows=181 width=16) (actual time=0.007..0.007 rows=1 loops=1)
          ->  Seq Scan on tmp2  (cost=0.00..34.45 rows=181 width=16) (actual time=0.002..0.003 rows=1 loops=1)
                Filter: ((b < 30) AND (d > 1))
  Total runtime: 0.063 ms

(9 rows)

SQL执行计划的分析：

1) 全表扫描左表TMP1，同时根据TMP1表子查询条件"C">=1和where过滤条件"T1"."B"<50联合过滤，即filter("TMP1"."B"<50 AND "C">=1)，计算结果临时表记为tmp1；

2) 全表扫描右表TMP2，同时根据TMP2表子查询条件"B"(+)<30和on子句"T2"."D"(+)>1联合过滤，即filter("TMP2"."D"(+)>1 AND "B"(+)<30)，计算结果临时表记为tmp2；

3) 左表TMP1及右表TMP2处理后临时表tmp1和tmp2通过access("TMP1"."A"="TMP2"."A"(+))连接条件进行Hash Left Join操作，左临时表结果集全量返回，右表不匹配行置为null，返回结果临时表记为tmp3；

4) 返回结果集。

2、SQL示例2：

SQL> select *

from (

select * from tmp1 where c >= 1

) t1 left join (

select * from tmp2 where b < 30

) t2 on t1.a = t2.a

and t2.d > 1 and t1.e >= 2

where t1.b < 50 and t2.e <= 3

;

A        B           C      E         A        B       D          E

---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------
      2       20           2      2         2           20       2          2

Execution Plan

----------------------------------------------------------

Plan hash value: 1630095649

---------------------------------------------------------------------------

| Id  | Operation       | Name | Rows  | Bytes | Cost (%CPU)| Time      |

---------------------------------------------------------------------------

|   0 | SELECT STATEMENT   |      |    1 |   104 |    7  (15)| 00:00:01 |

|*  1 |  HASH JOIN       |      |    1 |   104 |    7  (15)| 00:00:01 |

|*  2 |   TABLE ACCESS FULL| TMP2 |    1 |    52 |    3   (0)| 00:00:01 |

|*  3 |   TABLE ACCESS FULL| TMP1 |    3 |   156 |    3   (0)| 00:00:01 |

---------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

1 - access("TMP1"."A"="TMP2"."A")
    2 - filter("TMP2"."E"<=3 AND "TMP2"."D">1 AND "B"<30)
    3 - filter("TMP1"."B"<50 AND "TMP1"."E">=2 AND "C">=1)

Note

-----
    - dynamic sampling used for this statement (level=2)

Statistics

----------------------------------------------------------
       9  recursive calls
       0  db block gets
      15  consistent gets
       0  physical reads
       0  redo size
     981  bytes sent via SQL*Net to client
     524  bytes received via SQL*Net from client
       2  SQL*Net roundtrips to/from client
       0  sorts (memory)
       0  sorts (disk)
       1  rows processed

SQL>

postgres=# select *

postgres-# from (

postgres(# select * from tmp1 where c >= 1

postgres(# ) t1 left join (

postgres(# select * from tmp2 where b < 30

postgres(# ) t2 on t1.a = t2.a

postgres-# and t2.d > 1 and t1.e >= 2

postgres-# where t1.b < 50 and t2.e <= 3

postgres-# ;
  a | b  | c | e | a | b  | d | e

---+----+---+---+---+----+---+---
  2 | 20 | 2 | 2 | 2 | 20 | 2 | 2

(1 row)

postgres=# explain analyze select *

postgres-# from (

postgres(# select * from tmp1 where c >= 1

postgres(# ) t1 left join (

postgres(# select * from tmp2 where b < 30

postgres(# ) t2 on t1.a = t2.a

postgres-# and t2.d > 1 and t1.e >= 2

postgres-# where t1.b < 50 and t2.e <= 3

postgres-# ;
                                                  QUERY PLAN                                                 

-------------------------------------------------------------------------------------------------------------
  Hash Join  (cost=38.68..78.43 rows=18 width=32) (actual time=0.033..0.041 rows=1 loops=1)
    Hash Cond: ("outer".a = "inner".a)
    ->  Seq Scan on tmp1  (cost=0.00..38.53 rows=60 width=16) (actual time=0.007..0.011 rows=3 loops=1)
          Filter: ((c >= 1) AND (e >= 2) AND (b < 50))
    ->  Hash  (cost=38.53..38.53 rows=60 width=16) (actual time=0.008..0.008 rows=1 loops=1)
          ->  Seq Scan on tmp2  (cost=0.00..38.53 rows=60 width=16) (actual time=0.003..0.005 rows=1 loops=1)
                Filter: ((b < 30) AND (d > 1) AND (e <= 3))
  Total runtime: 0.070 ms

(8 rows)

postgres=#

SQL执行计划的分析：

1) 全表扫描左表TMP2，同时根据TMP2表子查询条件"B"<30和where过滤条件"TMP2"."E"<=3及ON子句过滤条件"TMP2"."D">1联合过滤，即filter("TMP2"."E"<=3 AND "TMP2"."D">1 AND "B"<30)，计算结果临时表记为tmp1；

2) 全表扫描右表TMP1，同时根据TMP1表子查询条件"C">=1和where子句过滤条件"TMP1"."B"<50及ON子句"TMP1"."E">=2联合过滤，即filter("TMP1"."B"<50 AND "TMP1"."E">=2 AND "C">=1)，计算结果临时表记为tmp2；

3) 临时表tmp1和tmp2通过access("TMP1"."A"="TMP2"."A")连接条件进行Hash Join连接操作（此处left join写法已经被转换为内链接），返回匹配结果临时表记为tmp3；

4) 返回结果集。