Handling skills of SQL optimization (2)

SQL Optimize , The old driver knows the handling skills （2）

Let's continue with the previous article

Replace subquery with join query

• mysql If you need to query data from two tables , There are generally two ways ： Subquery and join query , The subquery is as follows ：

• select * from order where id in(select * from user where status = 1);
  

• Subquery statements can be through in Keyword implementation , The condition of one query statement falls on another select Statement in the query result . The program first runs the statements nested in the innermost layer , Running an outer statement , The advantage of subquery statements is that they are simple , structured , If there are not many tables involved, we can use this method , But the subquery needs to create a temporary table , When the query is finished , You need to delete the temporary table again , There are some additional performance costs , At this time, we can change to join query

• select * from order inner join user on o.id = u.id and  status = 1;
  

join Don't have too many watches

• According to Alibaba developer manual ,join The number of tables should not exceed three , If join Too many tables for ,mysql Selecting an index can be very complicated , It's easy to make the wrong choice , If you don't hit the index , It will be very slow , We should try to reduce join Table number
• If there is no way to avoid too many in the actual business scenario join Table number , Then do it

join Pay attention to

• When we are involved in the joint query of multiple tables , You usually use join keyword , The best use is left join and inner join

• left join, When a join query is made between two tables , All the rows in the left table are returned , Even if there is no matching record in the right table

• right join, When a join query is made between two tables , All rows of the right table are returned , Even if there is no matching record in the left table

• inner join（ Internal connection ）, When a join query is made between two tables , Keep only the exact matching result sets in both tables

Control the number of indexes

• as everyone knows , Indexing can significantly improve query performance sql Performance of , But the number of indexes is not the more the better , Because when adding data to the table , You need to create an index for him at the same time , The index requires additional storage space , And there will be some performance consumption
• Alibaba stipulated in the development manual that , The number of indexes should not exceed 5 individual ,MySQL Use b+ The tree holds the index , During addition, deletion and modification , There will be additional performance consumption
• If the number of indexes required in the table exceeds 5 What should I do ？
• We can create a federated index without creating a single index , Useless indexes can be deleted , If the amount of data is huge , Let's consider not using mysql, have access to hbase

Select a reasonable field type

• char Represents a fixed string type , This type of field has a fixed amount of storage space , It wastes storage space

• alter table order add column code char(20) not null;
  

• varchar Represents variable length string type , The storage space of this type of field will be adjusted according to the length of the actual data , No waste of storage space

• alter table order add column code varchar(20) not null;
  

• If a field of fixed length , For example, the mobile phone numbers of users are generally 11 Bit , Then we directly define it as 11 Just a bit , However, if it is a field such as comments, we cannot use char, Waste a lot of space , We should use varchar

• When we choose the type , This principle should be followed ：

  • Can use numeric type , You don't need string types , Because string processing is slower than numbers
  • Use as small a type as possible
  • Type with fixed length , Such as char type
  • Variable length string type , Such as varchar type
  • Amount to be used decimal, Avoid loss of accuracy

promote group by The efficiency of

• We have a lot of scenarios that use group by keyword , Its main function is de duplication and grouping , Usually with having Use it together , It means that data is filtered according to certain conditions after grouping , Counterexample ：

• select * from order group by id having id < 200;
  

• This writing method is not good , He first looked up all the data and then grouped them , Then filter users id Less than 200 Users of , Grouping is a relatively time-consuming operation , Why can't we narrow it down first ？

• select * from order where id < 200 group by id ;
  

• Before grouping, we use where Conditions for filtration , Filter out the extra data , In this way, the efficiency of grouping will be improved , Before we do something time-consuming , Reduce the data range as much as possible

Index optimization

• sql Optimization , There is a very important content is ： Index optimization
• A lot of times , our sql sentence , Index gone and index not gone , Execution efficiency varies greatly , Index optimization is known as sql The first choice for optimization

sql Optimize

• Check sql The statement has no index

• explain select * from order where id < 200 group by id ;
  

• adopt explain Keyword we can see sql Whether or not to go , And then about explain Please search the Internet , Have time to explain