Question

有没有更好的方法来编写以下查询以优化性能。

INSERT INTO [dbo].[MBQ_All] ([Uniq], [StoreClass], [Store], [code], [ExtendedDescription], [ITEM_CLASS], [SUPPLIER], [Brands], [Min], [Max],[ADS]
           ,[Store+whMBQ])

    SELECT [Uniq], [StoreClass], [Store], [code], [ExtendedDescription], [ITEM_CLASS], [SUPPLIER], [Brands], [Min], [Max],[ADS]
           ,[Store+whMBQ]
    FROM dbo.MBQ_All_1
    WHERE MBQ_All_1.Uniq NOT IN (
            SELECT UNIQ
            FROM dbo.MBQ_All
            );

表格 MBQ_All 有超过4,00,000行，而 MBQ_All_1 有2,00,000行。

我正在使用SQL Server 2008。

Answer 1

也许尝试使用NOT EXISTS：

INSERT INTO [dbo].[MBQ_All] ([Uniq], [StoreClass], [Store], [code], [ExtendedDescription], [ITEM_CLASS], [SUPPLIER], [Brands], [Min], [Max],[ADS]
       ,[Store+whMBQ])

SELECT [Uniq], [StoreClass], [Store], [code], [ExtendedDescription], 
[ITEM_CLASS], [SUPPLIER], [Brands], [Min], [Max],[ADS] ,[Store+whMBQ]
FROM dbo.MBQ_All_1
WHERE NOT EXISTS (SELECT *
              FROM dbo.MBQ_All
              WHERE dbo.MBQ_All.UNIQ = MBQ_All_1.Uniq)

Answer 2

MERGE陈述的明显案例：

MERGE [dbo].[MBQ_All] Tgt
USING [dbo].[MBQ_All_1] Src ON Tgt.Uniq = Src.Uniq
WHEN NOT MATCHED THEN
    INSERT (    [Uniq],     [StoreClass],     [Store],     [code],     [ExtendedDescription],     [ITEM_CLASS],     [SUPPLIER],     [Brands],     [Min],     [Max],     [ADS],     [Store+whMBQ])
    VALUES (Src.[Uniq], Src.[StoreClass], Src.[Store], Src.[code], Src.[ExtendedDescription], Src.[ITEM_CLASS], Src.[SUPPLIER], Src.[Brands], Src.[Min], Src.[Max], Src.[ADS], Src.[Store+whMBQ]);

Answer 3

其中一个选择是LEFT JOIN，但我不确定它是否会更快：

INSERT INTO [dbo].[MBQ_All] 
     ([Uniq], [StoreClass], [Store], [code], [ExtendedDescription], [ITEM_CLASS], [SUPPLIER], [Brands], [Min], [Max],[ADS]               ,[Store+whMBQ])

    SELECT [Uniq], [StoreClass], [Store], [code], [ExtendedDescription], [ITEM_CLASS], [SUPPLIER], [Brands], [Min], [Max],[ADS]
           ,[Store+whMBQ]
    FROM dbo.MBQ_All_1 mbqN
    LEFT JOIN MBQ_All mbqO ON mbqN.Uniq =  mbqO.Uniq
    WHERE mbqO.Uniq IS NULL

查询优化，c＃中的超时错误

3 个答案: