同一张大表的高效连接

Question

我想加入一个带有大结果集（+ 6M 行）的表。

我有下表

Period	Zone	TAG	NAME	PROGRAM	COST
2020-10-01	EMEA	A	User2	Program 2	98
2020-11-01	APAC	A	User2	Program 2	103
2020-12-01	NA	A	User2	Program 2	118
2020-10-01	NA	X	User3	Program 3	334
2020-11-01	EMEA	X	User3	Program 3	362
2020-12-01	APAC	X	User3	Program 3	380

我想要一个具有这种输出的高效查询（我不介意没有值可以是 0 或 null）：

Period	Zone	TAG	NAME	PROGRAM	COST	COST EMEA	COST APAC	COST NA
2020-10-01	EMEA	A	User2	Program 2	98	98	0	0
2020-11-01	APAC	A	User2	Program 2	103	0	103	0
2020-12-01	NA	A	User2	Program 2	118	0	0	118
2020-10-01	NA	X	User3	Program 3	334	0	0	334
2020-11-01	EMEA	X	User3	Program 3	362	362	0	0
2020-12-01	APAC	X	User3	Program 3	380	0	380	0

如果我执行以下查询需要很长时间：

WITH TOTAL AS (
SELECT Period,Zone,TAG,NAME,PROGRAM,COST
FROM "DB"."SCHEMA"."Table"
)
SELECT T.*,E."COST" AS "COST EMEA",A."COST" AS "COST APAC",N."COST" AS "COST NA"
FROM TOTAL T , TOTAL E, TOTAL A, TOTAL N
WHERE T."Period" = E."Period" 
AND T."Period" = A."Period"
AND T."Period" = N."Period" 

我该如何改进这个查询？有没有一种有效的方法来执行它？

Answer 1

您的查询将需要很长时间才能运行，因为它正在对周期 ID 相同的行进行笛卡尔积 3 次（3 次连接）。我认为它也会产生错误的结果。

如果我正确理解您的问题，我认为您正在寻找这样的东西：

WITH TOTAL AS (
    SELECT
        Period,
        Zone,
        TAG,
        NAME,
        PROGRAM,
        COST
    FROM test_table
)
SELECT
    total.*,
    iff(upper(zone) = 'EMEA', cost, 0) as COST_EMEA,
    iff(upper(zone) = 'APAC', cost, 0) as COST_APAC,
    iff(upper(zone) = 'NA', cost, 0) as COST_NA
from total
;

这会生成下表（我认为您的示例结果不正确。表的最后一行将成本与NA 相关联，它应该是APAC。）：

PERIOD	ZONE	TAG	NAME	PROGRAM	COST	COST_EMEA	COST_APAC	COST_NA
2020-10-01	EMEA	A	User2	Program 2	98	98	0	0
2020-11-01	APAC	A	User2	Program 2	103	0	103	0
2020-12-01	NA	A	User2	Program 2	118	0	0	118
2020-10-01	NA	X	User3	Program 3	334	0	0	334
2020-11-01	EMEA	X	User3	Program 3	362	362	0	0
2020-12-01	APAC	X	User3	Program 3	380	0	380	0