Question

这是我的工作，虽然有效，但速度很慢。我需要对其进行优化。

SELECT main_table.id
FROM main_table
INNER JOIN linked_table latest_linked_record ON latest_linked_record.id = (
       SELECT clever_alias.id
       FROM linked_table clever_alias
       WHERE clever_alias.claim_order_id = main_table.id
       ORDER BY clever_alias.updated_at DESC,
                clever_alias.id DESC
       LIMIT 1
)
WHERE latest_linked_record.some_column NOT IN (1, 19)

似乎我一生都在与之抗争，与MySQL，MSSQL，Oracle和Postgres一样。除了物化视图之类的策略之外，我还能做些什么来改善性能？

Answer 1

在Postgres中，我建议使用DISTINCT ON：

SELECT mt.id
FROM main_table mt INNER JOIN
     (SELECT DISTINCT ON (lt.claim_order_id) lt.*
      FROM linked_table lt
      ORDER BY lt.claim_order_id, lt.claim_order_id DESC, lt.id DESC
     ) lt
     ON lt.claim_order_id = mt.id
WHERE lt.some_column NOT IN (1, 19);

当然，对于此查询，JOIN不是必需的：

SELECT lt.claim_order_id
FROM (SELECT DISTINCT ON (lt.claim_order_id) lt.*
      FROM linked_table lt
      ORDER BY lt.claim_order_id, lt.claim_order_id DESC, lt.id DESC
     ) lt
WHERE lt.some_column NOT IN (1, 19);

如何有效地从相关表中选择最新记录

1 个答案: