提高sql查询的性能

我正在寻找一些提示/技巧来提高插入到表中的多个SELECT语句的存储过程的性能。我加入的所有对象都已编入索引。提高sql查询的性能

我相信这个存储过程需要近一个小时运行的原因是因为有多个SELECT语句中使用以下两种观点：rvw_FinancialLineItemValues和rvw_FinancialLineItems

而且，每个SELECT语句使用特定的硬编码值的账户号码， LineItemTypeID以及来自上述两个视图的其他字段值。

如果我创建一个临时表，它可以一次获取这些SELECT语句所需的所有数据，然后在我的连接中使用此临时表，而它会提高性能吗？

是否有其他方法可以提高性能和可管理性？

 SELECT 
     @scenarioid, 
     @portfolioid, 
     pa.Id, 
     pa.ExternalID,  
     (select value from fn_split(i.AccountNumber,'.') where id = 1), 
     ac.[Description], 
     cl.Name, 
     NullIf((select value from fn_split(i.AccountNumber,'.') where id = 2),''), 
     NullIf((select value from fn_split(i.AccountNumber,'.') where id = 3),''), 
     ty.Name, 
     v.[Date], 
     cast(SUM(v.Amount) as decimal(13,2)), 
     GETDATE() 

    FROM rvw_FinancialLineItems i 
    INNER JOIN rvw_Scenarios sc 
     ON i.ScenarioId = sc.Id 
     AND sc.Id = @scenarioid 
     AND sc.PortfolioId = @portfolioid 
    INNER JOIN #pa AS pa 
     ON i.PropertyAssetID = pa.Id 
    INNER JOIN rvw_FinancialLineItemValues v 
     ON i.ScenarioId = v.ScenarioId 
     AND i.PropertyAssetID = v.PropertyAssetID 
     AND i.Id = v.FinancialLineItemId 
     AND ((i.BusinessEntityTypeId = 11 
     AND i.LineItemTypeId = 3002) 
     OR (i.LineItemTypeId IN (2005, 2010, 2003, 2125, 2209, 5012, 6001) 
     AND i.ModeledEntityKey = 1)) 
     AND i.AccountNumber not in ('401ZZ','403ZZ') 
    AND i.AccountNumber not in ('401XX') 
     AND i.AccountNumber not in ('40310','41110','42010','41510','40190','40110') -- exclude lease-level revenues selected below 
     AND v.[Date] BETWEEN @fromdate AND 
      CASE 
      WHEN pa.AnalysisEnd < @todate THEN pa.AnalysisEnd 
      ELSE @todate 
      END 
     AND v.ResultSet IN (0, 4) 
    INNER JOIN rvw_Portfolios po 
     ON po.Id = @portfolioid 
    INNER JOIN Accounts ac 
     ON po.ChartOfAccountId = ac.ChartOfAccountId 
     AND i.AccountNumber = ac.AccountNumber 
     AND ac.HasSubAccounts = 0 
    INNER JOIN fn_LookupClassTypes() cl 
     ON ac.ClassTypeId = cl.Id 
    INNER JOIN LineItemTypes ty 
     ON ac.LineItemTypeId = ty.Id 
    LEFT JOIN OtherRevenues r 
     ON i.PropertyAssetID = r.PropertyAssetID 
     AND i.AccountNumber = r.AccountID 
     AND v.[Date] BETWEEN r.[Begin] AND r.[End] 
    WHERE (r.IsMemo IS NULL 
    OR r.IsMemo = 0) 
    GROUP BY pa.AnalysisBegin 
       ,pa.Id 
       ,pa.ExternalID 
       ,i.AccountNumber 
       ,ac.[Description] 
       ,cl.Name 
       ,ty.Name 
       ,v.[Date] 
    HAVING SUM(v.amount) <> 0

来源

2016-12-14 NonProgrammer

您是否查看过查询执行计划和/或估计是否存在导致大部分执行时间的特定节点？ – Anand

首先，您使用的是UDF fn_split()？如果您没有使用表内值为的内联UDF，那么这是非常缓慢的。

二，是UDF fn_LookupClassTypes()一个内联表值UDF？如果不是，则将其转换为内联表值UDF。

最后，你的SQL查询有一些冗余。试试这个，看看它做了什么。

SELECT @scenarioid, @portfolioid, pa.Id, pa.ExternalID,  
     (select value from fn_split(i.AccountNumber,'.') 
     where id = 1), ac.[Description], cl.Name, 
    NullIf((select value from fn_split(i.AccountNumber,'.') 
      where id = 2),''), 
    NullIf((select value from fn_split(i.AccountNumber,'.') 
      where id = 3),''), ty.Name, v.[Date], 
    cast(SUM(v.Amount) as decimal(13,2)), GETDATE() 

FROM rvw_FinancialLineItems i 
    JOIN rvw_Scenarios sc ON sc.Id = i.ScenarioId 
    JOIN #pa AS pa ON pa.Id = i.PropertyAssetID 
    JOIN rvw_FinancialLineItemValues v 
     ON v.ScenarioId = i.ScenarioId 
     AND v.PropertyAssetID = i.PropertyAssetID 
     AND v.FinancialLineItemId = i.Id 
    JOIN rvw_Portfolios po ON po.Id = sc.portfolioid 
    JOIN Accounts ac 
     ON ac.ChartOfAccountId = po.ChartOfAccountId 
     AND ac.AccountNumber = i.AccountNumber 
    JOIN fn_LookupClassTypes() cl On cl.Id = ac.ClassTypeId 
    JOIN LineItemTypes ty On ty.Id = ac.LineItemTypeId 
    Left JOIN OtherRevenues r 
     ON r.PropertyAssetID = i.PropertyAssetID 
     AND r.AccountID = i.AccountNumber 
     AND v.[Date] BETWEEN r.[Begin] AND r.[End] 

WHERE i.ScenarioId = @scenarioid 
    and ac.HasSubAccounts = 0 
    and sc.PortfolioId = @portfolioid 
    and IsNull(r.IsMemo, 0) = 0) 
    and v.ResultSet In (0, 4) 
    and i.AccountNumber not in 
      ('401XX', '401ZZ','403ZZ','40310','41110', 
      '42010','41510','40190','40110') 
    and v.[Date] BETWEEN @fromdate AND 
      CASE WHEN pa.AnalysisEnd < @todate 
       THEN pa.AnalysisEnd ELSE @todate END 
    and ((i.LineItemTypeId = 3002 and i.BusinessEntityTypeId = 11) OR 
      (i.ModeledEntityKey = 1 and i.LineItemTypeId IN 
       (2005, 2010, 2003, 2125, 2209, 5012, 6001))) 

GROUP BY pa.AnalysisBegin,pa.Id, pa.ExternalID, i.AccountNumber, 
     ac.[Description],cl.Name,ty.Name,v.[Date] 
HAVING SUM(v.amount) <> 0

来源

2016-12-14 16:53:16