一起去大厂系列针对left join以及limit的两条优化小技巧

Posted 2021-10-16 来老铁干了这碗代码

记两则亲身经历的sql优化技巧：

一、主表数据不到100万，以下查询结果集约200左右，第一条sql执行效率为40ms，第二条为200ms，使用上面方法，其效率明显优于left join：

sql1：

SELECT SQL_NO_CACHE 
	usersr_id 
	,businessunit_id
	,ifnull((SELECT name FROM sync_businessunit WHERE id= obj.businessunit_id),'无名' ) businessunit_name
	,ifnull((SELECT fullname FROM sync_usersys WHERE id= obj.usersr_id),'无名' ) sr_name
	,SUM(price) price
	,IFNULL((SELECT target FROM analysis_target WHERE year=2017 and month=4 and usersr_id = obj.usersr_id  ),0) target
FROM analysis_cusorder	obj
WHERE send_time >= '2017-02-01' AND send_time < '2017-03-01' #BETWEEN '2017-02-01' and '2017-02-28' 
GROUP BY usersr_id 
ORDER BY usersr_id;

sql2：

SELECT SQL_NO_CACHE 
	o.usersr_id,
  o.businessunit_id,
  u.`name` as businessunit_name,
  us.fullname,
  t.target as zongzhibiao,
	SUM(o.price) as zongdacheng
FROM analysis_cusorder o
LEFT JOIN analysis_target t ON o.usersr_id = t.usersr_id and t.`year`=2017 and t.`month` = 4
LEFT JOIN sync_businessunit u ON o.businessunit_id = u.id
LEFT JOIN sync_usersys us ON o.usersr_id = us.id 
WHERE o.send_time BETWEEN '2017-02-01' and '2017-02-28' GROUP BY o.usersr_id;

---

二、为针对limit的优化，一般表数据超过1000万，limit基本就废了，需采用sql1的方法进行优化，效率相关极为明显，以下语句为使用php框架后的写法：

sql1：

$query = $this->db->select('id,third_id,recommend_menus')->
where("id > $maxid and recommend_menus != ''")->
order_by("id asc")->
limit($perpage)->
get('crawler_merchant');

sql2：

$query = $this->db->select('id,third_id,recommend_menus')->
limit($perpage, $offset)->
get('crawler_merchant');