Sql 如何加快子查询的速度，即从具有2000万行的表中选择数据

这是我的问题。此查询将导致20-30秒，并且在我的应用程序中导致超时错误。我试图用子句重写，但无法运行

对不起，我的英语很差

SELECT 
    StationName, StationID,
    (SELECT TOP (1) ISNULL(SUM(Liter), 0) AS TotalLiter 
     FROM Transactions  (nolock) 
     WHERE (StationID = s.StationID) 
       AND (SaleDate>= CONVERT(date, GETDATE())) 
       AND (FuelTypeID = 1)) AS Total1,
    (SELECT TOP (1) ISNULL(SUM(Liter), 0) AS TotalLiter
     FROM dbo.Transactions (nolock) AS Transactions_11 
     WHERE (StationID = s.StationID) 
       AND (SaleDate >= CONVERT(date, GETDATE())) 
       AND (FuelTypeID = 2)) AS Total2,
    (SELECT TOP (1) ISNULL(SUM(Liter), 0) AS TotalLiter
     FROM dbo.Transactions (nolock) AS Transactions_10
     WHERE (StationID = s.StationID) 
       AND (SaleDate >= CONVERT(date, GETDATE())) 
       AND (FuelTypeID = 3)) AS Total3,
    (SELECT TOP (1) ISNULL(SUM(Liter), 0) AS TotalLiter
     FROM dbo.Transactions (nolock) AS Transactions_9
     WHERE (StationID = s.StationID) 
       AND (SaleDate >= CONVERT(date, GETDATE())) 
       AND (FuelTypeID = 4)) AS Total4,
    (SELECT TOP (1) ISNULL(SUM(Liter), 0) AS TotalLiter
     FROM dbo.Transactions (nolock) AS Transactions_8
     WHERE (StationID = s.StationID) 
       AND (SaleDate >= CONVERT(date, GETDATE())) 
       AND (FuelTypeID = 5)) AS Total5,
                  (SELECT TOP (1) ISNULL(SUM(Liter),0) AS TotalLiter
                   FROM      dbo.Transactions  (nolock) AS Transactions_7
                   WHERE   (StationID = s.StationID) AND (SaleDate >= CONVERT(date, GETDATE())) AND (FuelTypeID = 6)) AS Total6,
                  (SELECT TOP (1) ISNULL(SUM(Liter),0) AS TotalLiter
                   FROM      dbo.Transactions  (nolock) AS Transactions_6
                   WHERE   (StationID = s.StationID) AND (SaleDate >= CONVERT(date, GETDATE())) AND (FuelTypeID = 7)) AS Total7,
                  (SELECT TOP (1) ISNULL(SUM(Liter),0) AS TotalLiter
                   FROM      dbo.Transactions  (nolock) AS Transactions_5
                   WHERE   (StationID = s.StationID) AND (SaleDate >= CONVERT(date, GETDATE())) AND (FuelTypeID = 8)) AS Total8,
                  (SELECT TOP (1) ISNULL(SUM(Liter),0) AS TotalLiter
                   FROM      dbo.Transactions  (nolock) AS Transactions_4
                   WHERE   (StationID = s.StationID) AND (SaleDate >= CONVERT(date, GETDATE())) AND (FuelTypeID = 9)) AS Total9,
                  (SELECT TOP (1) ISNULL(SUM(Liter),0) AS TotalLiter
                   FROM      dbo.Transactions  (nolock) AS Transactions_3
                   WHERE   (StationID = s.StationID) AND (SaleDate >= CONVERT(date, GETDATE())) AND (FuelTypeID = 10)) AS Total10,
                  (SELECT TOP (1) ISNULL(SUM(Liter),0) AS TotalLiter
                   FROM      dbo.Transactions  (nolock) AS Transactions_2
                   WHERE   (StationID = s.StationID) AND (SaleDate >= CONVERT(date, GETDATE()))) AS GrandTotal,
                  (SELECT TOP (1) MAX(SaleDate) AS LastSaleDate
                   FROM      dbo.Transactions (nolock) AS Transactions_1
WHERE   (StationID = s.StationID) AND (SaleDate >= CONVERT(date, GETDATE()))
                   ORDER BY LastSaleDate) AS LastSaleDate 
FROM    
    dbo.Stations (nolock) AS s 
WHERE 
    s.StationCode BETWEEN 1000 AND 1100 
ORDER BY 
    StationID

---

您可以使用

外部应用

和

条件聚合

避免多次运行相同的

子查询

试试这样的

SELECT StationName,
       StationID,
       Total1,
       Total2,
       Total3,
       Total4
       ...
       GrandTotal,
       LastSaleDate
FROM   dbo.Stations (nolock) AS s
       OUTER apply (SELECT Sum(CASE WHEN t.FuelTypeID = 1 THEN Liter END) AS Total1,
                           Sum(CASE WHEN t.FuelTypeID = 2 THEN Liter END) AS Total2,
                           Sum(CASE WHEN t.FuelTypeID = 3 THEN Liter END) AS Total3,
                           Sum(CASE WHEN t.FuelTypeID = 4 THEN Liter END) AS Total4,
                           ... 
                           Sum(Liter) as GrandTotal,
                           Max(SaleDate) AS LastSaleDate
                    FROM   Transactions t
                    WHERE  t.StationID = s.StationID
                           AND t.SaleDate >= CONVERT(DATE, Getdate())) ou 
WHERE  s.StationCode BETWEEN 1000 AND 1100
ORDER  BY StationID 

---

也不要忘记阅读本文：

您可以使用

外部应用

和

条件聚合

避免多次运行相同的

子查询

试试这样的

SELECT StationName,
       StationID,
       Total1,
       Total2,
       Total3,
       Total4
       ...
       GrandTotal,
       LastSaleDate
FROM   dbo.Stations (nolock) AS s
       OUTER apply (SELECT Sum(CASE WHEN t.FuelTypeID = 1 THEN Liter END) AS Total1,
                           Sum(CASE WHEN t.FuelTypeID = 2 THEN Liter END) AS Total2,
                           Sum(CASE WHEN t.FuelTypeID = 3 THEN Liter END) AS Total3,
                           Sum(CASE WHEN t.FuelTypeID = 4 THEN Liter END) AS Total4,
                           ... 
                           Sum(Liter) as GrandTotal,
                           Max(SaleDate) AS LastSaleDate
                    FROM   Transactions t
                    WHERE  t.StationID = s.StationID
                           AND t.SaleDate >= CONVERT(DATE, Getdate())) ou 
WHERE  s.StationCode BETWEEN 1000 AND 1100
ORDER  BY StationID 

---

另外，不要忘记阅读本文：

添加表结构、示例数据和预期结果D表结构、示例数据和预期结果我知道nolock的不良影响，但我把它作为最后的手段。谢谢你的回复。你的方式比我的查询要好得多，也干净得多：）我知道nolock的不良影响，但我把它作为最后的手段。谢谢你的回复。您的方式比我的查询更好、更干净：）