postgresql - 由于子查询，查询速度极慢

Question

我正在尝试将一些数据移植到一个新结构中，但这让我很费时间。

查询如下:

INSERT INTO atemp_addresses (id, city, state, country_id)
SELECT id, (SELECT name from cities WHERE id=adr.city_id limit 1),
(SELECT name FROM states WHERE id=(SELECT state_id FROM cities WHERE id=adr.city_id limit 1) limit 1), (SELECT country_id FROM states WHERE id=(SELECT state_id FROM cities WHERE id=adr.city_id limit 1) limit 1), FROM addresses adr

因为数据保存在其他表中，我必须使用所有这些子查询，这大大降低了速度。有更快的方法吗？

Answer 1

这是您的查询:

INSERT INTO atemp_addresses (id, city, state, country_id)
    SELECT id,
           (SELECT name from cities WHERE id = adr.city_id limit 1),
           (SELECT name FROM states WHERE id = (SELECT state_id FROM cities WHERE id=adr.city_id limit 1) limit 1),
           (SELECT country_id FROM states WHERE id=(SELECT state_id FROM cities WHERE id=adr.city_id limit 1) limit 1)
    FROM addresses adr;

您反复使用没有ORDER BY 的LIMIT。这意味着您将获得一个任意匹配行，该行可以从一次调用更改为下一次调用。

等效的查询应该是:

INSERT INTO atemp_addresses (id, city, state, country_id)
    SELECT adr.id, c.name, s.name, s.country_id
    FROM addresses adr JOIN
         cities c
         ON adr.city_id = c.id JOIN
         states s
         ON c.state_id = s.id;

您对 LIMIT 的使用表明可能有多个城市具有相同的 id(数据库设计不佳的标志？)。如果是这样:

INSERT INTO atemp_addresses (id, city, state, country_id)
    SELECT DISTINCT ON (adr.id) adr.id, c.name, s.name, s.country_id
    FROM addresses adr JOIN
         cities c
         ON adr.city_id = c.id JOIN
         states s
         ON c.state_id = s.id 
    ORDER BY adr.id;