使用條件連接優化SQL查詢

Question

我需要運行查詢來比較兩個表並返回不在第二個表中的記錄。棘手的部分是兩個表之間的鏈接是有條件的。我有多個數據源輸入表2，然後輸入表1。我試過使用一個連接，但是這並不比我下面提到的要快。這些表格中有大約5萬條記錄。該查詢大約需要1.5分鐘才能完成。我希望能把它弄到幾秒鐘。表1和表2已經在這些字段上有索引。數據庫在兼容級別SQL2008上運行。

此查詢時間約1.5分鐘：

select * 
from Table1 t1 
where not exists (select * 
        from Table2 t2 
        where t2.seedID = case t2.SeedSource when 'SeedSource1' then t1.SeedSource1 
                 when 'SeedSource2' then t1.SeedSource2 
                 when 'SeedSource3' then t1.SeedSource3 
                 when 'SeedSource4' then t1.SeedSource4 
                 when 'SeedSource5' then t1.SeedSource5 end) 

此查詢需要5分鐘以上：

select d.* 
from Tabel1 t1 left join 
    Table2 t2 on t2.seedID = case t2.SeedSource when 'SeedSource1' then t1.SeedSource1 
               when 'SeedSource2' then t1.SeedSource2 
               when 'SeedSource3' then t1.SeedSource3 
               when 'SeedSource4' then t1.SeedSource4 
               when 'SeedSource5' then t1.SeedSource5 end 
where t2.seedID is NULL 

任何幫助將不勝感激。

Answer 1

其實他們是不一樣的查詢作爲左連接將返回多行在多場比賽

不存在是更好的方法

我希望SeedSource，SeedSource1-5和seedID被索引

select * 
from Table1 t1 
where not exists (select * 
        from Table2 t2 
        where t2.seedID = t1.SeedSource1 
        and t2.SeedSource = 'SeedSource1' 
        union all 
        select * 
        from Table2 t2 
        where t2.seedID = t1.SeedSource2 
        and t2.SeedSource = 'SeedSource2' 
        //... 
       ) 

也許

left join Table2 t2 
    on (t2.seedID = t1.SeedSource1 and t2.SeedSource = 'SeedSource1') 
    or (t2.seedID = t1.SeedSource2 and t2.SeedSource = 'SeedSource2') 
    // ... 

Answer 2

閱讀您的要求後，您的查詢看起來很好。 （這是最好的做法是使用Not Exists代替Not IN和Left Join）

你可以做一個小小的優化：

• 在WHERE條款使用not exists (select 1 ...代替not exists (select * ...沒有必要選擇所有列當你可以選擇一個常數。 （更好的性能）

參考

• https://explainextended.com/2009/09/15/not-in-vs-not-exists-vs-left-join-is-null-sql-server/

Answer 3

可能：

SELECT t1.* FROM table1 t1 LEFT JOIN table2 t2 on t1.seedID = t2.seedID 
WHERE t2.seedID is NULL AND t1.SeedSource IN ('SeedSource1','SeedSource2','SeedSource3','SeedSource4','SeedSource5')