优化 CTE 查询

Question

我有 15 个表的大连接。简化后看起来像这样

WITH CTE AS
(
  SELECT A.ID,
         B.K,
         X.Date,
         X.ID,
         ROW_NUMBER() OVER (PARTITION BY A.ID ORDER BY X.date ASC, x.id ASC) RN
  FROM A
  INNER JOIN B ON A.ID = B.ID        
  INNER JOIN C ON B.ID = C.ID    
  LEFT JOIN X ON (X.G = A.ID
                 OR X.H = B.ID
                 OR X.I = C.ID)
)
SELECT * 
FROM CTE
WHERE RN = 1

执行时间约为 1 秒，主要是因为 X 的最后一个 JOIN。为了避免重复，我添加了 ROW_NUMBER 部分并将其变成了 CTE。

我如何最好使用 ANSI SQL 优化它，因为它在 SQL-Server 和 Oracle 上是运行？

Answer 1

我明白了。像这样超快（0.1秒）：

  SELECT A.ID,
         B.K,
         X.DATE,
         X.ID,
  FROM A
  INNER JOIN B ON A.ID = B.ID        
  INNER JOIN C ON B.ID = C.ID    
  LEFT JOIN 
  (
     SELECT G, H, I, ID, DATE,
     ROW_NUMBER() OVER (PARTITION BY G ORDER BY DATE ASC, ID ASC) RN
     FROM X
  ) X ON RN = 1 AND (X.G = A.ID
                     OR X.H = B.ID
                     OR X.I = C.ID)

优化 CTE 查询

Optimize CTE query

sql

sql-server

oracle

common-table-expression