0

I have this fact table:

CREATE TABLE [dbo].[FACT_TABLE]
(
    [A_id] [int] NOT NULL,
    [B_id] [int] NOT NULL,
    [C_id] [int] NOT NULL,
    [D_id] [int] NOT NULL,
    [FACT_total] [float] NOT NULL,
    [FACT_average] [float] NOT NULL,

    CONSTRAINT [PK_FACT_TABLE] 
        PRIMARY KEY CLUSTERED ([A_id] ASC, [B_id] ASC,
                               [C_id] ASC, [D_id] ASC)
) ON [PRIMARY]

I have a T-SQL Script to load data into my fact table:

INSERT INTO [dbo].[FACT_TABLE] ([A_id], [B_id], [C_id], [company_id], [D_id],[FACT_total], [FACT_average])
    SELECT 
        b.A_id, c.B_id, d.C_id, e.D_id,
        AVG(FACT_total) AS FACT_total,
        AVG(FACT_average) AS FACT_average
    FROM 
        [dbo].[staging_Area] a
    INNER JOIN 
        [dbo].[DIM_A] b ON a.A_id = b.A_id
    INNER JOIN 
        [dbo].[DIM_B] c ON a.[B_id] = c.B_id
    INNER JOIN 
        [dbo].[DIM_C] d ON a.[C_id] = d.C_id 
    INNER JOIN 
        [dbo].[DIM_D] e ON a.D_id = e.D_id
    GROUP BY  
        b.A_id, c.B_id, d.C_id, e.D_id

How can I guarantee that I don't insert duplicate values in my fact table in order to not get referential integrity error?

I tried with the following statement after that query:

EXCEPT 
SELECT *
FROM [dbo].[FACT_TABLE]

But I think that isn't the best approach for this...

Anyone have another alternative for this? Sorry but I can't put the real column and tables names...

Thanks!

4

2 回答 2

1

LEFT OUTER JOIN与WHERE FT.Column IS NULL事实表一起确保您没有插入已经存在的记录：

INSERT INTO [dbo].[FACT_TABLE] ([A_id],[B_id],[C_id],[company_id],[D_id],[FACT_total],[FACT_average])
SELECT b.A_id
        ,c.B_id
        ,d.C_id
        ,e.D_id
        ,AVG(FACT_total) AS FACT_total
        ,AVG(FACT_average) AS FACT_average
FROM [dbo].[staging_Area] a
INNER JOIN [dbo].[DIM_A] b ON
    a.A_id = b.A_id
INNER JOIN [dbo].[DIM_B] c ON
    a.[B_id] = c.B_id
INNER JOIN [dbo].[DIM_C] d ON
    a.[C_id] = d.C_id 
INNER JOIN [dbo].[DIM_D] e ON
    a.D_id = e.D_id
LEFT OUTER JOIN dbo.[FACT_TABLE] FT
    ON b.A_ID = FT.A_id
    AND c.B_id = FT.B_id
    AND D.C_ID = FT.C_id
    AND E.D_id = FT.D_id
WHERE FT.A_ID IS NULL
GROUP BY  b.A_id
        ,c.B_id
        ,d.C_id
        ,e.D_id
于 2017-07-11T18:21:32.750 回答
0

插入 [dbo].[FACT_TABLE]

（[A_id]、[B_id]、[C_id]、[D_id]、[FACT_total]、[FACT_average]）

选择 a.A_id、a.B_id、a.C_id、a.D_id、a.FACT_total、a.FACT_average

从

（ 选择

b.A_id, c.B_id, d.C_id, e.D_id,
AVG(FACT_total) AS FACT_total,
AVG(FACT_average) AS FACT_average

从

[dbo].[staging_Area] a

内部联接

[dbo].[DIM_A] b ON a.A_id = b.A_id

内部联接

[dbo].[DIM_B] c ON a.[B_id] = c.B_id

内部连接 ​​[dbo].[DIM_C] d ON a.[C_id] = d.C_id

内部联接

[dbo].[DIM_D] e ON a.D_id = e.D_id

通过...分组 

b.A_id, c.B_id, d.C_id, e.D_id 

)a

在哪里

不存在

（从 dbao.FACT_TABLE B 中选择 1

其中B.A_id
= A.A_id AND B.B_id = A.B_id 和 b.C_id=a.C_id 和 b.D_id=a.D_id

   )
于 2017-07-12T09:39:32.627 回答