Question

下面是我的示例查询，由于OPOS，POS4表中有很多记录（例如超过1000万条），因此执行时间很长。 where子句中的or条件对我造成了问题。

这是我的查询

SelecT DISTINCT T0.SaleID,T0.ParentID,T0.CustomerID from OPOS T0 
Left JOIN OCRD C0 ON C0.CustomerID = T0.ParentID --SS / Dist (FOR CUST TYPE) 
LEFT Join POS4 C1 ON T0.SaleID = C1.SaleID AND T0.ParentID = C1.ParentID AND C1.Type = 'C' AND C1.LevelType = 4 -- SS
LEFT Join POS4 C2 ON T0.SaleID = C2.SaleID AND T0.ParentID = C2.ParentID AND C2.Type = 'C' AND C2.LevelType = 2 -- DIST
LEFT Join POS4 C3 ON T0.SaleID = C3.SaleID AND T0.ParentID = C3.ParentID AND C3.Type = 'C' AND C3.LevelType = 3 -- DEALER
LEFT Join POS4 E1 ON T0.SaleID = E1.SaleID AND T0.ParentID = E1.ParentID AND E1.Type = 'E' -- LOGIN EMP USER(CUSTOMER // COMPANY)
LEFT Join POS4 E2 ON T0.SaleID = E2.SaleID AND T0.ParentID = E2.ParentID AND E2.Type = 'E' -- SELECTED EMP USER (COMPANY) 
Where Convert(Date,T0.Date) >= @FromDate AND Convert(Date,T0.Date) <= @ToDate AND C0.Type = @SaleBy AND T0.OrderType IN (12, 13) 
AND E1.TypeID = @LUserID AND E2.TypeID = @SUserID
AND (@SSID = 0 OR C1.TypeID = @SSID) 
AND (@DistributorID = 0 OR C2.TypeID = @DistributorID) 
AND (@DealerID = 0 OR C3.TypeID = @DealerID)

Answer 1

查看结果字段，您只需要T0的结果，而联接实际上是“内部联接”（在WHERE中隐含）。而不是内部（或左侧）连接所有这些表，简单的EXISTS检查可能会更好。如果您这样重写SQL怎么办？：

SelecT DISTINCT T0.SaleID,T0.ParentID,T0.CustomerID 
from OPOS T0
-- Why should you ever need a conversion of a date field to date?
-- not using Date but Text?
Where T0.[Date] >= @FromDate AND T0.[Date] < @ToDate  
AND EXISTS
(select * from OCRD C0 
  where C0.CustomerID = T0.ParentID and 
  C0.Type = @SaleBy AND T0.OrderType IN (12, 13)) --SS / Dist (FOR CUST TYPE)
AND EXISTS
(select * from POS4 E1 
  where T0.SaleID = E1.SaleID
  AND T0.ParentID = E1.ParentID 
  AND E1.Type = 'E'
  AND E1.TypeID = @LUserID) -- LOGIN EMP USER(CUSTOMER // COMPANY) 
AND EXISTS
(Select * from POS4 E2 
 where T0.SaleID = E2.SaleID 
   AND T0.ParentID = E2.ParentID 
   AND E2.Type = 'E'
   AND E2.TypeID = @SUserID) 
) -- SELECTED EMP USER (COMPANY) 

AND (@SSID = 0 OR 
 EXISTS
 (select * from POS4 C1 
  where T0.SaleID = C1.SaleID AND 
        T0.ParentID = C1.ParentID AND 
        C1.Type = 'C' AND 
        C1.LevelType = 4 AND
        C1.TypeID = @SSID)
) --SS
AND (@DistributorID = 0 OR 
  EXISTS
  ( select * from POS4 C2 
     where T0.SaleID = C2.SaleID 
       AND T0.ParentID = C2.ParentID 
       AND C2.Type = 'C' 
       AND C2.LevelType = 2 
       AND C2.TypeID = @DistributorID)
  )  -- DIST
AND (@DealerID = 0 OR 
      EXISTS
      (select * from POS4 C3 
       where T0.SaleID = C3.SaleID 
         AND T0.ParentID = C3.ParentID 
         AND C3.Type = 'C' 
         AND C3.LevelType = 3 
         AND C3.TypeID = @DealerID)
); -- DEALER

编辑：在代码中编辑了检查日期的where部分。

Where T0.[Date] >= @FromDate AND T0.[Date] < @ToDate

这将受益于[日期]上的索引。还要注意，它使用> =和<作为运算符（不是<= @ToDate）。这是在MS SQL Server中执行日期时间范围查询的正确方法（日期时间值具有3ms的敏感性）。您以这样的方式提供@FromDate和@ToDate值：@FromDate包含要包含的第一个DateTime，而@ToDate是要排除的最小DateTime。

转换为日期，您只想检查日期，并说要返回2018年6月的所有记录作为示例。那么您的@FromDate将为“ 20180601”（隐式转换为Datetime的2018年6月1日00:00:00），而@toDate将为“ 20180701”（隐式转换为2018年7月1日00:00:00）。这恰好意味着2018年6月的所有记录。哇，应该是：

t0.[Date] >= '20180601' and 
t0.[Date] <  '20180701'

完成后，它对应于：

Convert(date, t0.[Date]) >= '20180601' and 
Convert(date, t0.[Date]) <= '20180630'

但无需将Datetime转换为日期，并精确返回请求日期的数据。

在SQL查询中使用where-或条件的许多表中左联接

1 个答案: