如何優化此T-SQL查詢？

Question

我是相對較新的SQL，我希望對此有所幫助。基本上，我試圖在查詢中做一些計算，因爲函數有點慢。你們能改進嗎？

-- Query to retrive ADA from every school 
Select Distinct DY, 
     DATENAME(MM,DT) as 'Month', 
     CONVERT(nvarchar,DT,101) as 'Date', 
     (
      Select ((@M_stu - (Select Count(SC) from dbo.ATT where sc=1 AND al!='' AND DY=T.DY))/@M_stu)*100 
     ) as 'Merced ADA', 
     (
      Select ((@A_stu - (Select Count(SC) from dbo.ATT where sc=2 AND al!='' AND DY=T.DY))/@A_stu)*100 
     ) as 'Atwater ADA', 
     (
      Select ((@L_stu - (Select Count(SC) from dbo.ATT where sc=3 AND al!='' AND DY=T.DY))/@L_stu)*100 
     ) as 'Livingston ADA', 
     (
      Select ((@B_stu - (Select Count(SC) from dbo.ATT where sc=4 AND al!='' AND DY=T.DY))/@B_stu)*100 
     ) as 'Buhach ADA', 
     (
      Select ((@Y_stu - (Select Count(SC) from dbo.ATT where sc=5 AND al!='' AND DY=T.DY))/@Y_stu)*100 
     ) as 'Yosemite ADA', 
     (
      Select ((@I_stu - (Select Count(SC) from dbo.ATT where sc=6 AND al!='' AND DY=T.DY))/@I_stu)*100 
     ) as 'Independence ADA', 
     (
      Select ((@G_stu - (Select Count(SC) from dbo.ATT where sc=10 AND al!='' AND DY=T.DY))/@G_stu)*100 
     ) as 'Golden Valley ADA', 
     (
      Select ((@S_stu - (Select Count(SC) from dbo.ATT where sc=92 AND al!='' AND DY=T.DY))/@S_stu)*100 
     ) as 'Sequoia ADA' 
From dbo.ATT as T 
Order by DY ASC 

Answer 1

試試這個

Select DY, 
     DATENAME(MM,DT) as 'Month', 
     CONVERT(nvarchar,DT,101) as 'Date', 
     @M_stu - (Count(case when sc=1 AND al!='' AND DY=T.DY then 1 else null end)/@M_stu)*100 as 'Merced ADA', 
     @A_stu - (Count(case when sc=2 AND al!='' AND DY=T.DY then 1 else null end)/@A_stu)*100 as 'Atwater ADA', 
     ... 
From dbo.ATT as T 
GROUP BY DY, DT 
Order by DY ASC 

Answer 2

嘗試這種解決方案：

SELECT DISTINCT 
     t.DY       AS DY, 
     DATENAME(MM,t.DT)    AS [Month], 
     CONVERT(NVARCHAR,t.DT,101)  AS [Date], 
     ((@M_stu - d.[1])/@M_stu)*100 AS [Merced ADA], 
     ((@A_stu - d.[2])/@A_stu)*100 AS [Atwater ADA], 
     ((@L_stu - d.[3])/@L_stu)*100 AS [Livingston ADA], 
     ((@B_stu - d.[4])/@B_stu)*100 AS [Buhach ADA], 
     ((@Y_stu - d.[5])/@Y_stu)*100 AS [Yosemite ADA], 
     ((@I_stu - d.[6])/@I_stu)*100 AS [Independence ADA], 
     ((@G_stu - d.[10])/@G_stu)*100 AS [Golden Valley ADA], 
     ((@S_stu - d.[92])/@S_stu)*100 AS [Sequoia ADA] 
FROM dbo.ATT AS t 
OUTER APPLY (
    SELECT c.* 
    FROM (
     SELECT a.sc 
     FROM dbo.ATT AS a 
     WHERE a.sc IN (1,2,3,4,5,6,10,92) 
     AND  a.al!='' 
     AND  a.DY=t.DY 
    ) b 
    PIVOT(COUNT(b.sc) FOR a.sc IN ([1],[2],[3],[4],[5],[6],[10],[92])) c 
) d 
ORDER BY t.DY ASC; 

您也應該檢查，如果你有以下指標之一：CREATE [UNIQUE] INDEX index_name ON dbo.ATT (DY, sc, al)或CREATE [UNIQUE] INDEX index_name ON dbo.ATT (DY, sc) INCLUDE (al)。

Answer 3

你必須比較您當前查詢的結果，但是這是我怎麼可能會做到這一點：

Select Distinct DY, 
     DATENAME(MM,DT) as Month, 
     CONVERT(char(10),DT,101) as Date, 
     @M_stu - sum(case when sc=1 then 1.0 else 0.0 end)/@M_stu)*100 as [Merced ADA], 
     @A_stu - sum(case when sc=2 then 1.0 else 0.0 end)/@A_stu)*100 as [Atwater ADA], 
     @L_stu - sum(case when sc=3 then 1.0 else 0.0 end)/@L_stu)*100 as [Livingston ADA], 
     @B_stu - sum(case when sc=4 then 1.0 else 0.0 end)/@B_stu)*100 as [Buhach ADA], 
     @Y_stu - sum(case when sc=5 then 1.0 else 0.0 end)/@Y_stu)*100 as [Yosemite ADA], 
     @I_stu - sum(case when sc=6 then 1.0 else 0.0 end)/@I_stu)*100 as [Independence ADA], 
     @G_stu - sum(case when sc=10 then 1.0 else 0.0 end)/@G_stu)*100 as [Golden Valley ADA], 
     @S_stu - sum(case when sc=92 then 1.0 else 0.0 end)/@S_stu)*100 as [Sequoia ADA] 
From dbo.ATT as T 
Where al!='' 
Order by DY ASC 

而不必樣本數據來進行測試，我認爲這將是最簡單的方法採取。