Question

我有以下查询：

    SELECT 
    COUNT(DISTINCT CASE WHEN ResultsA.Followup = 1 THEN ResultsA.Auditors END) as ColumnResults,
    COUNT(DISTINCT CASE WHEN ResultsA.UnannInspectionYN = 1 THEN ResultsA.Auditors END) ColumnResults2,
    COUNT(DISTINCT CASE WHEN ResultsA.UnannInspectionYN = 0 AND ResultsA.Followup = 0 THEN ResultsA.Auditors END) ColumnResults3,
    COUNT(DISTINCT ResultsA.Auditors) ColumnResults4
FROM 
    ResultsA
    INNER JOIN TableB ON TableB.Auditors = ResultsA.Auditors 
    INNER JOIN TableC ON TableC.TableCColumn = ResultsA.TableCColumn AND Country IN (@Country)
WHERE 
exists(Select SupData 
                  from IMP_SUP_DATA 
                  where impID = @id and TableC.TableCColumn = SupData)
AND
    TableB.ColumnA in ( 'Accepted and reviewed' , 'Accepted, but not reviewed', 'Submitted', 'Accepted (provisional)')
AND 
    TableB.Date BETWEEN @FromDate AND @ToDate
AND 
    ISNULL(TableC.Datainfo,'No Data') IN (@Datainfo)

查询来自大量数据，但是当我在SSMS中运行它时，它的速度非常快。我可以回溯几年，查询最多仍会在约10秒内返回。但是，当我将查询作为报告运行时，在某些结果上可能要花几分钟甚至超时。

我该如何以最佳方式解决此问题？我希望从中学习，以防止将来发生。环境是sql server 2014。

Answer 1

OP在评论中提到，问题与SSMS和SSRS中的连接的不同SET选项无关

首先，通过运行报告并查询目录数据库来确认问题出在数据检索操作中

USE ReportServer
SELECT TOP 100 ReportPath, TimeDataRetrieval, TimeProcessing, TimeRendering 
FROM dbo.ExecutionLog2
ORDER BY TimeStart desc

TimeDataRetrieval的值应占整个持续时间的大部分

如果是这种情况，并且具有相同参数的相同查询在SSMS中运行很快，而在SSRS中运行缓慢，请考虑以下更改：

将变量的显式CAST添加为确切类型，作为表中相应列的类型
在WHERE子句中删除ISNULL（）
添加OPTION RECOMPILE，因为这是参数化的资源密集型查询

要评估以下查询（但首先要对CAST进行调整以适合正确的数据类型）：

SELECT 
    COUNT(DISTINCT CASE WHEN ResultsA.Followup = 1 THEN ResultsA.Auditors END) as ColumnResults,
    COUNT(DISTINCT CASE WHEN ResultsA.UnannInspectionYN = 1 THEN ResultsA.Auditors END) ColumnResults2,
    COUNT(DISTINCT CASE WHEN ResultsA.UnannInspectionYN = 0 AND ResultsA.Followup = 0 THEN ResultsA.Auditors END) ColumnResults3,
    COUNT(DISTINCT ResultsA.Auditors) ColumnResults4
FROM 
    ResultsA
    INNER JOIN TableB ON TableB.Auditors = ResultsA.Auditors 
    INNER JOIN TableC ON TableC.TableCColumn = ResultsA.TableCColumn AND Country IN (CAST(@Country AS VARCHAR(255)) )
WHERE 
EXISTS (Select SupData 
                  from IMP_SUP_DATA 
                  where impID = CAST(@id AS INT) and TableC.TableCColumn = SupData)

AND    TableB.ColumnA in ( 'Accepted and reviewed' , 'Accepted, but not reviewed', 'Submitted', 'Accepted (provisional)')

AND    TableB.Date BETWEEN CAST(@FromDate AS DATETIME) AND CAST(@ToDate AS DATETIME)
AND    ( TableC.Datainfo IN (CAST(@Datainfo AS VARCHAR(255))) OR ( CAST(@Datainfo AS VARCHAR(255))  = 'No Data' AND TableC.Datainfo IS NULL))
OPTION ( RECOMPILE)

SSRS的查询速度慢，但SSMS的查询速度快

1 个答案: