Question

我有3个表：t1，t2，t3。

t1 - 分层表（具有id和parent_id）

t1与t2连接，t2与t3（连接）

连接

输入：t1的一些id

有必要在这些标识符上构建一个树（以收集所有父节点）

然后构建一棵树，将t2和t3粘在一起

还需要恢复其子女的t3name（来自t3）的值为父母的上衣

输出：必须为每个t3name接收分层树

我的SQL请求，但速度非常慢：

Miniexample： http://sqlfiddle.com/#!4/0b607/2

SELECT id, name, t3name FROM 
(SELECT distinct t1.id as id, 
                 t1.parent as parent, 
           t1.name as name, 
           connect_by_root(t3.name) as t3name
    FROM table1 t1
    LEFT OUTER JOIN table2 t2
        ON t1.t2_id = t2.id
    LEFT OUTER JOIN table3 t3
        ON t2.t3_id = t3.id
    START WITH t1.id in (83, 98, 9, 19, 4, 101)
        CONNECT BY PRIOR t1.parent = t1.id)
WHERE t3name IS NOT NULL
START WITH parent = 0
    CONNECT BY PRIOR id = parent and PRIOR t3name = t3name

Answer 1

在没有看到执行计划的情况下帮助您调整查询有点困难，但如果我理解您，那么您只需要t3name的值为2个连接。

如果确实如此，那么首先尝试从table1获取分层数据并将结果连接到table2和table3。

试试这样：

SELECT id, name, t3name FROM 
(SELECT distinct t1.id as id, 
                 t1.parent as parent, 
           t1.name as name, 
           t3.name as t3name
    FROM (select tt1.name, tt1.id, tt1.parent, connect_by_root(tt1.t2_id) t2_id
         from table1 tt1 
         START WITH tt1.id in (83, 98, 9, 19, 4, 101)
        CONNECT BY PRIOR tt1.parent = tt1.id) t1
    JOIN table2 t2
        ON t1.t2_id = t2.id
    JOIN table3 t3
        ON t2.t3_id = t3.id
    )
WHERE t3name IS NOT NULL
START WITH parent = 0
    CONNECT BY PRIOR id = parent and PRIOR t3name = t3name;

我向你的小提琴添加了一些主键（看看我是否可以让它使用索引）并放入两个查询。见here

如何从2 CONNECT BY提高SQL查询的速度

1 个答案: