Question

我正在我的存储过程中执行一个数据透视操作，如下所示，但是当我将它用于报告时，它会影响性能。

是否有任何方式/替代此枢轴操作以提高性能？或者我可以为此申请索引吗？

SELECT colId,colSequence,colDescription,
ISNULL([1],0) AS 'IESO',ISNULL([2],0) AS 'RFRC',ISNULL([3],0) AS 'EAL',ISNULL([4],0) AS 'HNS',ISNULL([5],0) AS 'PTE',ISNULL([6],0) AS 'EOC',ISNULL([7],0) AS 'MU'
FROM  
(  
 SELECT PF.FeatureId,PF.Sequence,PF.[Description],PTB.BenefitId,
 FROM PSP.table1 PF 
 LEFT JOIN PSP.table2 PTB  ON PF.FeatureId = PTB.FeatureId
 left JOIN PSP.table3 PO  ON PO.productID = PF.productId
 WHERE PF.ProductId = @ProductId  AND PF.IsDeleted = 0

) PS  
PIVOT  
(  
MAX (BenefitId)  
FOR BenefitId IN  
( [1],[2],[3],[4],[5],[6],[7])  
) AS pvt  
ORDER BY colSequence

Answer 1

您的查询使用left JOIN PSP.table3 PO，但此表格不会显示在子查询的SELECT字段列表中。

如果此表无效，您可以将其删除，以便获得操作。

检查查询表上的索引，但我认为您FeatureId上的索引与productID字段上的索引

相同

Answer 2

不确定它是否超级有益，但您可以使用聚合案例表达式来获取所需的值，而不是PIVOT。您可以使用OUTER APPLY，这样您就不必按3列分组，其中一列是描述列。

SELECT  colId           = PF.FeatureId,
        colSequence     = PF.Sequence,
        colDescription  = PF.[Description],
        PTB.*
FROM    PSP.table1 PF
        OUTER APPLY (
            SELECT  'IESO'  = COUNT(CASE WHEN BenefitId = 1 THEN 1 END),
                    'RFRC'  = COUNT(CASE WHEN BenefitId = 2 THEN 1 END),
                    'EAL'   = COUNT(CASE WHEN BenefitId = 3 THEN 1 END),
                    'HNS'   = COUNT(CASE WHEN BenefitId = 4 THEN 1 END),
                    'PTE'   = COUNT(CASE WHEN BenefitId = 5 THEN 1 END),
                    'EOC'   = COUNT(CASE WHEN BenefitId = 6 THEN 1 END),
                    'MU'    = COUNT(CASE WHEN BenefitId = 7 THEN 1 END)
            FROM    PSP.table2 PTB 
            WHERE   PF.FeatureId = PTB.FeatureId
        ) PTB
WHERE   PF.ProductId = @ProductId
        AND PF.IsDeleted = 0

SQL - 改进数据透视查询性能

2 个答案: