Question

我有一个存储过程可以生成一个select语句（5000条记录）。在select语句中我调用了表值函数。表值函数执行一些聚合计算

我花了一分多钟才能返回结果

这是聚合函数代码

ALTER FUNCTION [dbo].[GetParticipationEvaluation](@Participation_ID int)
RETURNS @TabEvaluation TABLE (
   JudgeCount               int      NOT NULL,
   JudgeEvaluationCount     int     NOT NULL,
   Evaluation               float

) 
AS
BEGIN
   declare @JudgeCount int
   declare @JudgeEvaluationCount int
   declare @Evaluation float
   SELECT 
   @JudgeCount=COUNT(*),@Evaluation= AVG(Evaluation) from Judge_Participation where Participation_ID=@Participation_ID and Sys_Is_Active=1
    and status_ID =2
    SELECT 
   @JudgeEvaluationCount=COUNT(*),@Evaluation= AVG(Evaluation) from Judge_Participation where Participation_ID=@Participation_ID and Sys_Is_Active=1
    INSERT INTO @TabEvaluation (JudgeCount, JudgeEvaluationCount, Evaluation)
    select @JudgeCount,@JudgeEvaluationCount,@Evaluation
   RETURN;
END;

这是存储过程代码

ALTER procedure [dbo].[GetEvalutionStatistics]
as
select  [Participation].[Participation_ID]
    ,[Participation].[Name_Ar]
    ,[Participation].[Name_En]
    ,[Participation].[Is_Submitted]
    ,[Participation].[Student_ID]
    ,[Participation].[Result_ID]
    ,[Participation].[Field_ID]
    ,[Participation].[Category_ID]
    ,[Participation].[Actual_Field_ID]
    ,[Participation].[Current_Member_ID]
    ,[Participation].[Sys_Is_Active]
    ,[Participation].[Turnitin_Value]
    ,[Participation].[Turnitin_Link]
    ,Eval.JudgeCount
    ,Eval.JudgeEvaluationCount
    ,Eval.Evaluation
    ,[Actual_Field_ID].[Name_Ar] as 'Actual_Field_ID.Name_Ar'
    ,[Actual_Field_ID].[Name_En] as 'Actual_Field_ID.Name_En'
    ,[Result_ID].[Name_Ar] as 'Result_ID.Name_Ar'
    ,[Result_ID].[Name_En] as 'Result_ID.Name_En'
    --,dbo.getjudgecount(participation_ID,null) 'JudgeCount'
    --,dbo.getjudgecount(participation_ID,2) 'JudgeEvaluationCount'
    --,dbo.GetAvgParticipationEvaluation(Participation_ID) 'Evaluation'
    --,dbo.getParticipationSpecialist(Participation_ID) as 'specialist'
from [Participation] 
    left join [Participation_Field_List] as [Actual_Field_ID] on [Actual_Field_ID].[Field_ID]=[Participation].[Actual_Field_ID]
    left join [Participation_Result] as [Result_ID] on [Result_ID].[Result_ID]=[Participation].[Result_ID]
OUTER APPLY dbo.GetParticipationEvaluation([Participation].participation_ID) Eval
where 
     participation.Sys_Is_Active=1 and participation.is_submitted=1

我想知道造成这种不良表现的部分

Answer 1

通常你需要对'join columns'和'where子句'进行索引，尝试获取执行计划，看看是否缺少索引。

Answer 2

尝试不使用慢速功能的此过程

CREATE procedure [dbo].[GetEvalutionStatistics]
as
select  p.[Participation_ID]
    ,p.[Name_Ar]
    ,p.[Name_En]
    ,p.[Is_Submitted]
    ,p.[Student_ID]
    ,p.[Result_ID]
    ,p.[Field_ID]
    ,p.[Category_ID]
    ,p.[Actual_Field_ID]
    ,p.[Current_Member_ID]
    ,p.[Sys_Is_Active]
    ,p.[Turnitin_Value]
    ,p.[Turnitin_Link]
    ,Eval.JudgeCount
    ,Eval.JudgeEvaluationCount
    ,Eval.Evaluation
    ,[Actual_Field_ID].[Name_Ar] as 'Actual_Field_ID.Name_Ar'
    ,[Actual_Field_ID].[Name_En] as 'Actual_Field_ID.Name_En'
    ,[Result_ID].[Name_Ar] as 'Result_ID.Name_Ar'
    ,[Result_ID].[Name_En] as 'Result_ID.Name_En'
    --,dbo.getjudgecount(participation_ID,null) 'JudgeCount'
    --,dbo.getjudgecount(participation_ID,2) 'JudgeEvaluationCount'
    --,dbo.GetAvgParticipationEvaluation(Participation_ID) 'Evaluation'
    --,dbo.getParticipationSpecialist(Participation_ID) as 'specialist'
from [Participation] p
    left join [Participation_Field_List] as [Actual_Field_ID] on [Actual_Field_ID].[Field_ID]= p.[Actual_Field_ID]
    left join [Participation_Result] as [Result_ID] on [Result_ID].[Result_ID]= p.[Result_ID]
OUTER APPLY (
             SELECT COUNT(CASE WHEN Sys_Is_Active=1 and status_ID = 2 THEN 1 END) AS JudgeCount,
                    AVG(CASE WHEN Sys_Is_Active=1 THEN Evaluation END) AS Evaluation,
                    COUNT(CASE WHEN Sys_Is_Active=1 THEN 1 END) AS JudgeEvaluationCount      
             FROM Judge_Participation 
             WHERE Participation_ID = p.participation_ID and (Sys_Is_Active=1 OR status_ID =2)
             ) Eval
where p.Sys_Is_Active=1 and p.is_submitted=1

Answer 3

好吧，我发现你在外部应用中使用的函数中有一些逻辑。没有这个功能你测试过吗？当然，你将缺少逻辑，但要确定函数或查询的其余部分是否是问题，这是可能的。

除非我真的需要，否则我不会尝试使用函数。对于我所看到的，您可以将所有逻辑放入存储过程中。

我会查看此查询将生成的查询计划。如果您不知道如何做到这一点，可以在http://plixa.nl/how-to-read-query-execution-plans/找到解决方法。

Answer 4

我认为在此查询中使用函数导致性能成本。由于在查询中调用用户定义的函数，因此会降低性能。

所以请考虑尝试替代方法以避免功能。

请检查表格中的索引。

请参阅以下链接

http://www.mindfiresolutions.com/SqlServer-Query-Optimization-Tips-470.php

http://www.mssqltips.com/sql-server-tip-category/37/query-optimization/

http://blog.sqlauthority.com/2010/02/16/sql-server-stored-procedure-optimization-tips-best-practices/

使用聚合函数的SQL Server性能问题

4 个答案: