如何在 mysql 或 php 中优化或加速我的子查询

Question

示例查询：

 SELECT 
     table1.t1_id,table1.name,
     table2.address,
     (
         SELECT message 
         FROM table3 
         WHERE logid =  table1.t1_id 
               AND message NOT LIKE "[ SYSTEM%" 
         ORDER BY logs 
         DESC LIMIT 1
     ) as message   
 FROM table1 
 INNER JOIN table2 
 ON table1.t1_id = table2.t2_id   
 WHERE table1.dateCreated   
 BETWEEN CAST('2015-01-01' as Date) 
         AND CAST('2015-05-30' as Date)   
 ORDER BY table1.dateCreated DESC

预期输出：

id | name | address | message |

注意：假设表 1 和表 2 有数千行，而表 3 有数百万行

Answer 1

这是您查询的修改版本

select 
t1.t1_id,
t1.name,
t2.address,
t3.message
from table1 t1
join table2 t2 on t1.t1_id = t2.t2_id
join table3 t3 on t3.logid = t1.t1_id
join(
  select max(logs) as logs,logid from table3
  where 
  message NOT LIKE "[ SYSTEM%" 
  group by logid
)x
on x.logs = t3.logs and x.logid = t3.logid
where 
t1.dateCreated BETWEEN CAST('2015-01-01' as Date) AND CAST('2015-05-30' as Date)
ORDER BY t1.dateCreated DESC;

现在您将需要表上的索引，我假设所有表的 id 表是主键并被索引，因此无需在它们上添加额外的索引。 需要以下索引

alter table table3 add index logs_idx(logs);
alter table table3 add index message_idx(message);
alter table table3 add index logid_idx(logid);

alter table table1 add index dateCreated_idx(dateCreated);

注意：确保在应用索引之前进行表备份。

Answer 2

这样更快吗？

SELECT table1.t1_id, table1.name,table2.address, m.message 
FROM 
table3 as m, 
table1 INNER JOIN table2 ON table1.t1_id = table2.t2_id
WHERE table1.dateCreated
BETWEEN CAST('2015-01-01' as Date) AND CAST('2015-05-30' as Date)
and m.logid = table1.t1_id
and m.message not like "[ SYSTEM%"
having min(m.log)
ORDER BY table1.dateCreated DESC