Question

这适用于PostgreSQL数据库。我有一个观点，我将自己加入桌子15次以创建我们称之为关卡的级别，从那里我在＆＃39;级别上执行COALESCE功能 - 然后对该字段进行一些操作，以及。我也在拉15个级别中的每个级别的描述字段。这是我的查询变得缓慢的地方。我将SETHEADERT表连接到多个级别以获取每个级别的描述字段。如您所见，我只有3个描述字段，目前运行时间很长。当我有2时它花了一点点但不是很糟糕。我希望这是有道理的。我的代码如下。任何有关如何提高效率的帮助都非常感谢。

SELECT 

subset_cls,
prctr1,

CASE
WHEN prctr1 LIKE 'PC%' THEN split_part( overlay(prctr1 placing '00000' from 1 for 2 ),'.',1)
ELSE prctr1 end as pctrl2,

LVL01,
desc01,
LVL02,
desc02


FROM
( SELECT
SRC.SAP_SETNODE.SUBSET_CLS AS SUBSET_CLS,
SRC.SAP_SETHEADERT.DESCRIPTION AS desc01,
DESC_02.DESCRIPTION AS desc02,
DESC_03.DESCRIPTION AS desc03,
SRC.SAP_SETNODE.SET_NAME AS LVL01, 
SRC.SAP_SETNODE.SUBSET_NAME AS LVL02, 
SETNODE_1.SUBSET_NAME AS LVL03, 
SETNODE_2.SUBSET_NAME AS LVL04, 
SETNODE_3.SUBSET_NAME AS LVL05, 
SETNODE_4.SUBSET_NAME AS LVL06, 
SETNODE_5.SUBSET_NAME AS LVL07, 
SETNODE_6.SUBSET_NAME AS LVL08, 
SETNODE_7.SUBSET_NAME AS LVL09, 
SETNODE_8.SUBSET_NAME AS LVL10, 
SETNODE_9.SUBSET_NAME AS LVL11, 
SETNODE_10.SUBSET_NAME AS LVL12, 
SETNODE_11.SUBSET_NAME AS LVL13, 
SETNODE_12.SUBSET_NAME AS LVL14, 
SETNODE_13.SUBSET_NAME AS LVL15,
COALESCE(
SETNODE_13.SUBSET_NAME,
SETNODE_12.SUBSET_NAME,
SETNODE_11.SUBSET_NAME,
SETNODE_10.SUBSET_NAME,
SETNODE_9.SUBSET_NAME,
SETNODE_8.SUBSET_NAME,
SETNODE_7.SUBSET_NAME,
SETNODE_6.SUBSET_NAME,
SETNODE_5.SUBSET_NAME,
SETNODE_4.SUBSET_NAME,
SETNODE_3.SUBSET_NAME,
SETNODE_2.SUBSET_NAME,
SETNODE_1.SUBSET_NAME,
SRC.SAP_SETNODE.SUBSET_NAME,
SRC.SAP_SETNODE.SET_NAME)
AS prctr1 
FROM SRC.SAP_SETNODE
LEFT JOIN SRC.SAP_SETHEADERT  ON SRC.SAP_SETHEADERT.SET_NAME = SRC.SAP_SETNODE.SET_NAME

LEFT JOIN SRC.SAP_SETNODE AS SETNODE_1 ON SRC.SAP_SETNODE.SUBSET_NAME = SETNODE_1.SET_NAME
AND SRC.SAP_SETNODE.SUBSET_CLS = SETNODE_1.SUBSET_CLS

LEFT JOIN SRC.SAP_SETHEADERT as DESC_02  ON DESC_02.SET_NAME = SETNODE_1.SET_NAME

LEFT JOIN SRC.SAP_SETNODE AS SETNODE_2 ON SETNODE_1.SUBSET_NAME = SETNODE_2.SET_NAME 
AND SETNODE_1.SUBSET_CLS = SETNODE_2.SUBSET_CLS

LEFT JOIN SRC.SAP_SETHEADERT as DESC_03  ON DESC_03.SET_NAME = SETNODE_2.SET_NAME



LEFT JOIN SRC.SAP_SETNODE AS SETNODE_3 ON SETNODE_2.SUBSET_NAME = SETNODE_3.SET_NAME 
AND SETNODE_2.SUBSET_CLS = SETNODE_3.SUBSET_CLS



LEFT JOIN SRC.SAP_SETNODE AS SETNODE_4 ON SETNODE_3.SUBSET_NAME = SETNODE_4.SET_NAME 
AND SETNODE_3.SUBSET_CLS = SETNODE_4.SUBSET_CLS



LEFT JOIN SRC.SAP_SETNODE AS SETNODE_5 ON SETNODE_4.SUBSET_NAME = SETNODE_5.SET_NAME
AND SETNODE_4.SUBSET_CLS = SETNODE_5.SUBSET_CLS


LEFT JOIN SRC.SAP_SETNODE AS SETNODE_6 ON SETNODE_5.SUBSET_NAME = SETNODE_6.SET_NAME
AND SETNODE_5.SUBSET_CLS = SETNODE_6.SUBSET_CLS


LEFT JOIN SRC.SAP_SETNODE AS SETNODE_7 ON SETNODE_6.SUBSET_NAME = SETNODE_7.SET_NAME
AND SETNODE_6.SUBSET_CLS = SETNODE_7.SUBSET_CLS



LEFT JOIN SRC.SAP_SETNODE AS SETNODE_8 ON SETNODE_7.SUBSET_NAME = SETNODE_8.SET_NAME
AND SETNODE_7.SUBSET_CLS = SETNODE_8.SUBSET_CLS



LEFT JOIN SRC.SAP_SETNODE AS SETNODE_9 ON SETNODE_8.SUBSET_NAME = SETNODE_9.SET_NAME
AND SETNODE_8.SUBSET_CLS = SETNODE_9.SUBSET_CLS



LEFT JOIN SRC.SAP_SETNODE AS SETNODE_10 ON SETNODE_9.SUBSET_NAME = SETNODE_10.SET_NAME 
AND SETNODE_9.SUBSET_CLS = SETNODE_10.SUBSET_CLS 

LEFT JOIN SRC.SAP_SETNODE AS SETNODE_11 ON SETNODE_10.SUBSET_NAME = SETNODE_11.SET_NAME 
AND SETNODE_10.SUBSET_CLS = SETNODE_11.SUBSET_CLS

LEFT JOIN SRC.SAP_SETNODE AS SETNODE_12 ON SETNODE_11.SUBSET_NAME = SETNODE_12.SET_NAME
AND SETNODE_11.SUBSET_CLS = SETNODE_12.SUBSET_CLS

LEFT JOIN SRC.SAP_SETNODE AS SETNODE_13 ON SETNODE_12.SUBSET_NAME = SETNODE_13.SET_NAME
AND SETNODE_12.SUBSET_CLS = SETNODE_13.SUBSET_CLS

GROUP BY SRC.SAP_SETNODE.SUBSET_CLS, SRC.SAP_SETHEADERT.DESCRIPTION, DESC_02.DESCRIPTION,              
DESC_03.DESCRIPTION, SRC.SAP_SETNODE.SET_NAME, 
SRC.SAP_SETNODE.SUBSET_NAME, SETNODE_1.SUBSET_NAME, SETNODE_2.SUBSET_NAME, 
SETNODE_3.SUBSET_NAME, SETNODE_4.SUBSET_NAME, SETNODE_5.SUBSET_NAME, 
SETNODE_6.SUBSET_NAME, SETNODE_7.SUBSET_NAME, SETNODE_8.SUBSET_NAME, 
SETNODE_9.SUBSET_NAME, SETNODE_10.SUBSET_NAME, SETNODE_11.SUBSET_NAME, 
SETNODE_12.SUBSET_NAME, SETNODE_13.SUBSET_NAME

HAVING SRC.SAP_SETNODE.SUBSET_CLS='0101' AND SRC.SAP_SETNODE.SET_NAME='SISW.'  
||get_fy_part('YEAR',       clock_timestamp())

ORDER BY SRC.SAP_SETNODE.SET_NAME, SRC.SAP_SETNODE.SUBSET_NAME, SETNODE_1.SUBSET_NAME, 
SETNODE_2.SUBSET_NAME, SETNODE_3.SUBSET_NAME, SETNODE_4.SUBSET_NAME, SETNODE_5.SUBSET_NAME, 
SETNODE_6.SUBSET_NAME, SETNODE_7.SUBSET_NAME

) foo

Answer 1

如果将过滤移动到where子句，则要连接的行数将会减少。改变这个：

group by ...
HAVING
    SRC.SAP_SETNODE.SUBSET_CLS = '0101' AND
    SRC.SAP_SETNODE.SET_NAME = 'SISW.' || get_fy_part('YEAR', clock_timestamp())

到

where
    SRC.SAP_SETNODE.SUBSET_CLS = '0101' AND
    SRC.SAP_SETNODE.SET_NAME = 'SISW.' || get_fy_part('YEAR', clock_timestamp())
group by ...

如何让这个讨厌的查询更有效率？

1 个答案: