在连接条件中使用 OR JOIN 还是使用多个 JOINS 更好
Is it better to JOIN using OR in the join condition or to use multiple JOINS
我正在尝试根据票证是否表明某个事件已发生来标记票证。我 运行 遇到的问题是表明发生这种情况的特定值可能在票证上的 25 个字段中的任何一个中，而我当前的逻辑需要相当长的时间才能完成（通常 >45 分钟) 如果它完全完成（我们的 DBA 偶尔会因为占用太多 CPU 或 PJI 太高而将其杀死）。目前我正在使用以下方法来尝试标记票证：
CREATE VOLATILE TABLE
--DROP TABLE
FCT
AS (
SELECT
I_FCT.Ticket_Nbr
, I_FCT.Plan_Dim_CK
, I_FCT.S_DATE
, I_FCT.E_DATE
, CASE
WHEN I_FCT.Event1_CK = AD.Event_CK THEN '1'
WHEN I_FCT.Event2_CK = AD.Event_CK THEN '2'
WHEN I_FCT.Event3_CK = AD.Event_CK THEN '3'
WHEN I_FCT.Event4_CK = AD.Event_CK THEN '4'
WHEN I_FCT.Event5_CK = AD.Event_CK THEN '5'
WHEN I_FCT.Event6_CK = AD.Event_CK THEN '6'
WHEN I_FCT.Event7_CK = AD.Event_CK THEN '7'
WHEN I_FCT.Event8_CK = AD.Event_CK THEN '8'
WHEN I_FCT.Event9_CK = AD.Event_CK THEN '9'
WHEN I_FCT.Event10_CK = AD.Event_CK THEN '10'
WHEN I_FCT.Event11_CK = AD.Event_CK THEN '11'
WHEN I_FCT.Event12_CK = AD.Event_CK THEN '12'
WHEN I_FCT.Event13_CK = AD.Event_CK THEN '13'
WHEN I_FCT.Event14_CK = AD.Event_CK THEN '14'
WHEN I_FCT.Event15_CK = AD.Event_CK THEN '15'
/*WHEN I_FCT.Event16_CK = AD.Event_CK THEN '16'
WHEN I_FCT.Event17_CK = AD.Event_CK THEN '17'
WHEN I_FCT.Event18_CK = AD.Event_CK THEN '18'
WHEN I_FCT.Event19_CK = AD.Event_CK THEN '19'
WHEN I_FCT.Event20_CK = AD.Event_CK THEN '20'
WHEN I_FCT.Event21_CK = AD.Event_CK THEN '21'
WHEN I_FCT.Event22_CK = AD.Event_CK THEN '22'
WHEN I_FCT.Event23_CK = AD.Event_CK THEN '23'
WHEN I_FCT.Event24_CK = AD.Event_CK THEN '24'
WHEN I_FCT.Event25_CK = AD.Event_CK THEN '25'*/
ELSE NULL
END AS AOD_Event
FROM
I_FCT
LEFT JOIN AOD_CKS AS AD
ON (AD.Event_CK = I_FCT.Event1_CK
OR AD.Event_CK = I_FCT.Event2_CK
OR AD.Event_CK = I_FCT.Event3_CK
OR AD.Event_CK = I_FCT.Event4_CK
OR AD.Event_CK = I_FCT.Event5_CK
OR AD.Event_CK = I_FCT.Event6_CK
OR AD.Event_CK = I_FCT.Event7_CK
OR AD.Event_CK = I_FCT.Event8_CK
OR AD.Event_CK = I_FCT.Event9_CK
OR AD.Event_CK = I_FCT.Event10_CK
OR AD.Event_CK = I_FCT.Event11_CK
OR AD.Event_CK = I_FCT.Event12_CK
OR AD.Event_CK = I_FCT.Event13_CK
OR AD.Event_CK = I_FCT.Event14_CK
OR AD.Event_CK = I_FCT.Event15_CK
/*OR AD.Event_CK = I_FCT.Event16_CK
OR AD.Event_CK = I_FCT.Event17_CK
OR AD.Event_CK = I_FCT.Event18_CK
OR AD.Event_CK = I_FCT.Event19_CK
OR AD.Event_CK = I_FCT.Event20_CK
OR AD.Event_CK = I_FCT.Event21_CK
OR AD.Event_CK = I_FCT.Event22_CK
OR AD.Event_CK = I_FCT.Event23_CK
OR AD.Event_CK = I_FCT.Event24_CK
OR AD.Event_CK = I_FCT.Event25_CK*/
)
)
WITH DATA PRIMARY INDEX (Ticket_Nbr)
ON COMMIT PRESERVE ROWS;
所以我的问题是每个事件 CK 切换到一次 JOINing 是否会更好（顺便说一句，这是一个关键值，是 AOD_CKS 上的索引）？或者我应该不那么冗长并尝试拆分工作（即加入一次用于字段 1-6，另一个 table 用于 7-13，等等）然后将这 4 个 table 加入在一起？
非常感谢任何帮助！
我最终将 JOIN 拆分为 3 个部分，然后按如下方式合并数据：
----------------------------------------------------------------------------
-- FLAG ADDICTION CLAIMS
----------------------------------------------------------------------------
CREATE VOLATILE TABLE
--DROP TABLE
I_FCT_1
AS (
SELECT
I_FCT.Claim_Nbr
, I_FCT.Plan_Dim_CK
, CASE
WHEN I_FCT.Diag1_CK = AD.Diag_CK THEN '1'
WHEN I_FCT.Diag2_CK = AD.Diag_CK THEN '2'
WHEN I_FCT.Diag3_CK = AD.Diag_CK THEN '3'
WHEN I_FCT.Diag4_CK = AD.Diag_CK THEN '4'
WHEN I_FCT.Diag5_CK = AD.Diag_CK THEN '5'
WHEN I_FCT.Diag6_CK = AD.Diag_CK THEN '6'
WHEN I_FCT.Diag7_CK = AD.Diag_CK THEN '7'
WHEN I_FCT.Diag8_CK = AD.Diag_CK THEN '8'
ELSE NULL
END AS AOD_Diag
FROM
I_FCT
LEFT JOIN ADD_CKS AS AD
ON (AD.Diag_CK = I_FCT.Diag1_CK
OR AD.Diag_CK = I_FCT.Diag2_CK
OR AD.Diag_CK = I_FCT.Diag3_CK
OR AD.Diag_CK = I_FCT.Diag4_CK
OR AD.Diag_CK = I_FCT.Diag5_CK
OR AD.Diag_CK = I_FCT.Diag6_CK
OR AD.Diag_CK = I_FCT.Diag7_CK
OR AD.Diag_CK = I_FCT.Diag8_CK
)
)
WITH DATA PRIMARY INDEX (Claim_Nbr)
ON COMMIT PRESERVE ROWS;
----------------------------------------------------------------------------
CREATE VOLATILE TABLE
--DROP TABLE
I_FCT_2
AS (
SELECT
I_FCT.Claim_Nbr
, I_FCT.Plan_Dim_CK
, CASE
WHEN I_FCT.Diag9_CK = AD.Diag_CK THEN '9'
WHEN I_FCT.Diag10_CK = AD.Diag_CK THEN '10'
WHEN I_FCT.Diag11_CK = AD.Diag_CK THEN '11'
WHEN I_FCT.Diag12_CK = AD.Diag_CK THEN '12'
WHEN I_FCT.Diag13_CK = AD.Diag_CK THEN '13'
WHEN I_FCT.Diag14_CK = AD.Diag_CK THEN '14'
WHEN I_FCT.Diag15_CK = AD.Diag_CK THEN '15'
WHEN I_FCT.Diag16_CK = AD.Diag_CK THEN '16'
ELSE NULL
END AS AOD_Diag
FROM
I_FCT
LEFT JOIN ADD_CKS AS AD
ON (AD.Diag_CK = I_FCT.Diag9_CK
OR AD.Diag_CK = I_FCT.Diag10_CK
OR AD.Diag_CK = I_FCT.Diag11_CK
OR AD.Diag_CK = I_FCT.Diag12_CK
OR AD.Diag_CK = I_FCT.Diag13_CK
OR AD.Diag_CK = I_FCT.Diag14_CK
OR AD.Diag_CK = I_FCT.Diag15_CK
OR AD.Diag_CK = I_FCT.Diag16_CK
)
)
WITH DATA PRIMARY INDEX (Claim_Nbr)
ON COMMIT PRESERVE ROWS;
----------------------------------------------------------------------------
CREATE VOLATILE TABLE
--DROP TABLE
I_FCT_3
AS (
SELECT
I_FCT.Claim_Nbr
, I_FCT.Plan_Dim_CK
, CASE
WHEN I_FCT.Diag17_CK = AD.Diag_CK THEN '17'
WHEN I_FCT.Diag18_CK = AD.Diag_CK THEN '18'
WHEN I_FCT.Diag19_CK = AD.Diag_CK THEN '19'
WHEN I_FCT.Diag20_CK = AD.Diag_CK THEN '20'
WHEN I_FCT.Diag21_CK = AD.Diag_CK THEN '21'
WHEN I_FCT.Diag22_CK = AD.Diag_CK THEN '22'
WHEN I_FCT.Diag23_CK = AD.Diag_CK THEN '23'
WHEN I_FCT.Diag24_CK = AD.Diag_CK THEN '24'
WHEN I_FCT.Diag25_CK = AD.Diag_CK THEN '25'
ELSE NULL
END AS AOD_Diag
FROM
I_FCT
LEFT JOIN ADD_CKS AS AD
ON (AD.Diag_CK = I_FCT.Diag17_CK
OR AD.Diag_CK = I_FCT.Diag18_CK
OR AD.Diag_CK = I_FCT.Diag19_CK
OR AD.Diag_CK = I_FCT.Diag20_CK
OR AD.Diag_CK = I_FCT.Diag21_CK
OR AD.Diag_CK = I_FCT.Diag22_CK
OR AD.Diag_CK = I_FCT.Diag23_CK
OR AD.Diag_CK = I_FCT.Diag24_CK
OR AD.Diag_CK = I_FCT.Diag25_CK
)
)
WITH DATA PRIMARY INDEX (Claim_Nbr)
ON COMMIT PRESERVE ROWS;
---------------------------------------------------------------------------
CREATE VOLATILE TABLE
--DROP TABLE
FCT
AS (
SELECT
I_FCT.Claim_Nbr
, I_FCT.Plan_Dim_CK
, I_FCT.ADMISSION_DATE
, I_FCT.DISCHARGE_DATE
, I_FCT.ADMISSION_DATE_DIM_CK
, I_FCT.Proc3_CK
, I_FCT.Proc4_CK
, I_FCT.Proc5_CK
, I_FCT.Proc6_CK
, I_FCT.Proc7_CK
, I_FCT.Bill_Type
, CASE
WHEN C1.AOD_Diag IS NOT NULL THEN C1.AOD_Diag
WHEN C2.AOD_Diag IS NOT NULL THEN C2.AOD_Diag
WHEN C3.AOD_Diag IS NOT NULL THEN C3.AOD_Diag
ELSE NULL
END AS AOD_Diag
FROM
I_FCT
JOIN I_FCT_1 AS C1
ON C1.Claim_Nbr = I_FCT.Claim_Nbr
AND C1.Plan_Dim_CK = I_FCT.Plan_Dim_CK
JOIN I_FCT_2 AS C2
ON I_FCT.Claim_Nbr = C2.Claim_Nbr
AND I_FCT.Plan_Dim_CK = C2.Plan_Dim_CK
JOIN I_FCT_3 AS C3
ON I_FCT.Claim_Nbr = C3.Claim_Nbr
AND I_FCT.Plan_Dim_CK = C3.Plan_Dim_CK
)
WITH DATA PRIMARY INDEX (Claim_Nbr)
ON COMMIT PRESERVE ROWS;
我正在尝试根据票证是否表明某个事件已发生来标记票证。我 运行 遇到的问题是表明发生这种情况的特定值可能在票证上的 25 个字段中的任何一个中，而我当前的逻辑需要相当长的时间才能完成（通常 >45 分钟) 如果它完全完成（我们的 DBA 偶尔会因为占用太多 CPU 或 PJI 太高而将其杀死）。目前我正在使用以下方法来尝试标记票证：
CREATE VOLATILE TABLE
--DROP TABLE
FCT
AS (
SELECT
I_FCT.Ticket_Nbr
, I_FCT.Plan_Dim_CK
, I_FCT.S_DATE
, I_FCT.E_DATE
, CASE
WHEN I_FCT.Event1_CK = AD.Event_CK THEN '1'
WHEN I_FCT.Event2_CK = AD.Event_CK THEN '2'
WHEN I_FCT.Event3_CK = AD.Event_CK THEN '3'
WHEN I_FCT.Event4_CK = AD.Event_CK THEN '4'
WHEN I_FCT.Event5_CK = AD.Event_CK THEN '5'
WHEN I_FCT.Event6_CK = AD.Event_CK THEN '6'
WHEN I_FCT.Event7_CK = AD.Event_CK THEN '7'
WHEN I_FCT.Event8_CK = AD.Event_CK THEN '8'
WHEN I_FCT.Event9_CK = AD.Event_CK THEN '9'
WHEN I_FCT.Event10_CK = AD.Event_CK THEN '10'
WHEN I_FCT.Event11_CK = AD.Event_CK THEN '11'
WHEN I_FCT.Event12_CK = AD.Event_CK THEN '12'
WHEN I_FCT.Event13_CK = AD.Event_CK THEN '13'
WHEN I_FCT.Event14_CK = AD.Event_CK THEN '14'
WHEN I_FCT.Event15_CK = AD.Event_CK THEN '15'
/*WHEN I_FCT.Event16_CK = AD.Event_CK THEN '16'
WHEN I_FCT.Event17_CK = AD.Event_CK THEN '17'
WHEN I_FCT.Event18_CK = AD.Event_CK THEN '18'
WHEN I_FCT.Event19_CK = AD.Event_CK THEN '19'
WHEN I_FCT.Event20_CK = AD.Event_CK THEN '20'
WHEN I_FCT.Event21_CK = AD.Event_CK THEN '21'
WHEN I_FCT.Event22_CK = AD.Event_CK THEN '22'
WHEN I_FCT.Event23_CK = AD.Event_CK THEN '23'
WHEN I_FCT.Event24_CK = AD.Event_CK THEN '24'
WHEN I_FCT.Event25_CK = AD.Event_CK THEN '25'*/
ELSE NULL
END AS AOD_Event
FROM
I_FCT
LEFT JOIN AOD_CKS AS AD
ON (AD.Event_CK = I_FCT.Event1_CK
OR AD.Event_CK = I_FCT.Event2_CK
OR AD.Event_CK = I_FCT.Event3_CK
OR AD.Event_CK = I_FCT.Event4_CK
OR AD.Event_CK = I_FCT.Event5_CK
OR AD.Event_CK = I_FCT.Event6_CK
OR AD.Event_CK = I_FCT.Event7_CK
OR AD.Event_CK = I_FCT.Event8_CK
OR AD.Event_CK = I_FCT.Event9_CK
OR AD.Event_CK = I_FCT.Event10_CK
OR AD.Event_CK = I_FCT.Event11_CK
OR AD.Event_CK = I_FCT.Event12_CK
OR AD.Event_CK = I_FCT.Event13_CK
OR AD.Event_CK = I_FCT.Event14_CK
OR AD.Event_CK = I_FCT.Event15_CK
/*OR AD.Event_CK = I_FCT.Event16_CK
OR AD.Event_CK = I_FCT.Event17_CK
OR AD.Event_CK = I_FCT.Event18_CK
OR AD.Event_CK = I_FCT.Event19_CK
OR AD.Event_CK = I_FCT.Event20_CK
OR AD.Event_CK = I_FCT.Event21_CK
OR AD.Event_CK = I_FCT.Event22_CK
OR AD.Event_CK = I_FCT.Event23_CK
OR AD.Event_CK = I_FCT.Event24_CK
OR AD.Event_CK = I_FCT.Event25_CK*/
)
)
WITH DATA PRIMARY INDEX (Ticket_Nbr)
ON COMMIT PRESERVE ROWS;
所以我的问题是每个事件 CK 切换到一次 JOINing 是否会更好（顺便说一句，这是一个关键值，是 AOD_CKS 上的索引）？或者我应该不那么冗长并尝试拆分工作（即加入一次用于字段 1-6，另一个 table 用于 7-13，等等）然后将这 4 个 table 加入在一起？
非常感谢任何帮助！
我最终将 JOIN 拆分为 3 个部分，然后按如下方式合并数据：
----------------------------------------------------------------------------
-- FLAG ADDICTION CLAIMS
----------------------------------------------------------------------------
CREATE VOLATILE TABLE
--DROP TABLE
I_FCT_1
AS (
SELECT
I_FCT.Claim_Nbr
, I_FCT.Plan_Dim_CK
, CASE
WHEN I_FCT.Diag1_CK = AD.Diag_CK THEN '1'
WHEN I_FCT.Diag2_CK = AD.Diag_CK THEN '2'
WHEN I_FCT.Diag3_CK = AD.Diag_CK THEN '3'
WHEN I_FCT.Diag4_CK = AD.Diag_CK THEN '4'
WHEN I_FCT.Diag5_CK = AD.Diag_CK THEN '5'
WHEN I_FCT.Diag6_CK = AD.Diag_CK THEN '6'
WHEN I_FCT.Diag7_CK = AD.Diag_CK THEN '7'
WHEN I_FCT.Diag8_CK = AD.Diag_CK THEN '8'
ELSE NULL
END AS AOD_Diag
FROM
I_FCT
LEFT JOIN ADD_CKS AS AD
ON (AD.Diag_CK = I_FCT.Diag1_CK
OR AD.Diag_CK = I_FCT.Diag2_CK
OR AD.Diag_CK = I_FCT.Diag3_CK
OR AD.Diag_CK = I_FCT.Diag4_CK
OR AD.Diag_CK = I_FCT.Diag5_CK
OR AD.Diag_CK = I_FCT.Diag6_CK
OR AD.Diag_CK = I_FCT.Diag7_CK
OR AD.Diag_CK = I_FCT.Diag8_CK
)
)
WITH DATA PRIMARY INDEX (Claim_Nbr)
ON COMMIT PRESERVE ROWS;
----------------------------------------------------------------------------
CREATE VOLATILE TABLE
--DROP TABLE
I_FCT_2
AS (
SELECT
I_FCT.Claim_Nbr
, I_FCT.Plan_Dim_CK
, CASE
WHEN I_FCT.Diag9_CK = AD.Diag_CK THEN '9'
WHEN I_FCT.Diag10_CK = AD.Diag_CK THEN '10'
WHEN I_FCT.Diag11_CK = AD.Diag_CK THEN '11'
WHEN I_FCT.Diag12_CK = AD.Diag_CK THEN '12'
WHEN I_FCT.Diag13_CK = AD.Diag_CK THEN '13'
WHEN I_FCT.Diag14_CK = AD.Diag_CK THEN '14'
WHEN I_FCT.Diag15_CK = AD.Diag_CK THEN '15'
WHEN I_FCT.Diag16_CK = AD.Diag_CK THEN '16'
ELSE NULL
END AS AOD_Diag
FROM
I_FCT
LEFT JOIN ADD_CKS AS AD
ON (AD.Diag_CK = I_FCT.Diag9_CK
OR AD.Diag_CK = I_FCT.Diag10_CK
OR AD.Diag_CK = I_FCT.Diag11_CK
OR AD.Diag_CK = I_FCT.Diag12_CK
OR AD.Diag_CK = I_FCT.Diag13_CK
OR AD.Diag_CK = I_FCT.Diag14_CK
OR AD.Diag_CK = I_FCT.Diag15_CK
OR AD.Diag_CK = I_FCT.Diag16_CK
)
)
WITH DATA PRIMARY INDEX (Claim_Nbr)
ON COMMIT PRESERVE ROWS;
----------------------------------------------------------------------------
CREATE VOLATILE TABLE
--DROP TABLE
I_FCT_3
AS (
SELECT
I_FCT.Claim_Nbr
, I_FCT.Plan_Dim_CK
, CASE
WHEN I_FCT.Diag17_CK = AD.Diag_CK THEN '17'
WHEN I_FCT.Diag18_CK = AD.Diag_CK THEN '18'
WHEN I_FCT.Diag19_CK = AD.Diag_CK THEN '19'
WHEN I_FCT.Diag20_CK = AD.Diag_CK THEN '20'
WHEN I_FCT.Diag21_CK = AD.Diag_CK THEN '21'
WHEN I_FCT.Diag22_CK = AD.Diag_CK THEN '22'
WHEN I_FCT.Diag23_CK = AD.Diag_CK THEN '23'
WHEN I_FCT.Diag24_CK = AD.Diag_CK THEN '24'
WHEN I_FCT.Diag25_CK = AD.Diag_CK THEN '25'
ELSE NULL
END AS AOD_Diag
FROM
I_FCT
LEFT JOIN ADD_CKS AS AD
ON (AD.Diag_CK = I_FCT.Diag17_CK
OR AD.Diag_CK = I_FCT.Diag18_CK
OR AD.Diag_CK = I_FCT.Diag19_CK
OR AD.Diag_CK = I_FCT.Diag20_CK
OR AD.Diag_CK = I_FCT.Diag21_CK
OR AD.Diag_CK = I_FCT.Diag22_CK
OR AD.Diag_CK = I_FCT.Diag23_CK
OR AD.Diag_CK = I_FCT.Diag24_CK
OR AD.Diag_CK = I_FCT.Diag25_CK
)
)
WITH DATA PRIMARY INDEX (Claim_Nbr)
ON COMMIT PRESERVE ROWS;
---------------------------------------------------------------------------
CREATE VOLATILE TABLE
--DROP TABLE
FCT
AS (
SELECT
I_FCT.Claim_Nbr
, I_FCT.Plan_Dim_CK
, I_FCT.ADMISSION_DATE
, I_FCT.DISCHARGE_DATE
, I_FCT.ADMISSION_DATE_DIM_CK
, I_FCT.Proc3_CK
, I_FCT.Proc4_CK
, I_FCT.Proc5_CK
, I_FCT.Proc6_CK
, I_FCT.Proc7_CK
, I_FCT.Bill_Type
, CASE
WHEN C1.AOD_Diag IS NOT NULL THEN C1.AOD_Diag
WHEN C2.AOD_Diag IS NOT NULL THEN C2.AOD_Diag
WHEN C3.AOD_Diag IS NOT NULL THEN C3.AOD_Diag
ELSE NULL
END AS AOD_Diag
FROM
I_FCT
JOIN I_FCT_1 AS C1
ON C1.Claim_Nbr = I_FCT.Claim_Nbr
AND C1.Plan_Dim_CK = I_FCT.Plan_Dim_CK
JOIN I_FCT_2 AS C2
ON I_FCT.Claim_Nbr = C2.Claim_Nbr
AND I_FCT.Plan_Dim_CK = C2.Plan_Dim_CK
JOIN I_FCT_3 AS C3
ON I_FCT.Claim_Nbr = C3.Claim_Nbr
AND I_FCT.Plan_Dim_CK = C3.Plan_Dim_CK
)
WITH DATA PRIMARY INDEX (Claim_Nbr)
ON COMMIT PRESERVE ROWS;