Sql 存在多对多关系

我正在使用Amazon Redshift，无法正确获取查询。假设我有很多项目，每个项目都需要一个或多个技能。因此，项目和技能之间存在着多对多的关系。用户创建项目，因此每个项目有一个创建者

对于每个项目，我想让所有其他项目都由至少共享一项技能的同一个创建者完成。所以我想写这样的东西：

SELECT p1.project_id, p2.project_id
FROM projects p1 
JOIN projects p2 on p1.creator = p2.creator
WHERE EXISTS (SELECT 0
              from skills sk1, skills sk2 
              where sk1.project_id = p1.project_id 
                and sk2.project_id = p2.project_id 
                and sk1.skill = sk2.skill)

SELECT distinct p1.project_id, p2.project_id
   FROM projects p1
   JOIN projects p2 on 
      p1.creator = p2.creator and
      p2.project_id > p1.project_id
   join skills sk1 on 
      sk1.project_id = p1.project_id
   join skills sk2 on 
      sk2.project_id = p2.project_id and
      sk1.skill = sk2.skill

问题是这确实是（给出一个

磁盘已满

错误）

以下方法可行，但速度也很慢（需要半小时左右）：

这个问题是，如果我想在第二个项目上聚合一些属性，我必须将其用作子查询

---

有更好的方法吗？我原以为第一个查询会更快，因为它只限于挑选一个项目

查询中的一个简单问题是，您允许项目加入自身。这意味着每个项目都将被返回

通过确保连接的两个项目不相同来解决此问题：

JOIN projects p2 on 
   p1.creator = p2.creator and
   p2.project_id > p1.project_id

请注意，我使用的是

，而不是

=以便两个匹配的项目仅在一个方向上连接。否则，每对项目将返回两次

然后，基于加入的解决方案将如下所示：

SELECT p1.project_id, p2.project_id
FROM projects p1 
JOIN projects p2 on p1.creator = p2.creator
WHERE EXISTS (SELECT 0
              from skills sk1, skills sk2 
              where sk1.project_id = p1.project_id 
                and sk2.project_id = p2.project_id 
                and sk1.skill = sk2.skill)

SELECT distinct p1.project_id, p2.project_id
   FROM projects p1
   JOIN projects p2 on 
      p1.creator = p2.creator and
      p2.project_id > p1.project_id
   join skills sk1 on 
      sk1.project_id = p1.project_id
   join skills sk2 on 
      sk2.project_id = p2.project_id and
      sk1.skill = sk2.skill

在连接列上是否设置了索引？