19 MySQL optimizations commonly used in projects

Before writing the article , First, thank you. Feiyou Technology Documents provided by Mr. Lu ..

Make a statement ： The following optimization schemes are all based on “ Mysql- Indexes -BTree type ” Of

One 、EXPLAIN

do MySQL Optimize , We need to make good use of EXPLAIN see SQL Implementation plan .

Here's a simple example , mark (1,2,3,4,5) The data we need to focus on

1. type Column , Connection type . A good sql The statement must at least reach range Level . Put an end to it all Level
2. key Column , Index name used . If index is not selected , The value is NULL. You can use forced indexing
3. key_len Column , Index length
4. rows Column , Number of scanning lines . This is an estimate
5. extra Column , Detailed instructions . Note that the common unfriendly values are ：Using filesort, Using temporary

Two 、SQL In the sentence IN Should not contain too many values

MySQL about IN The corresponding optimization has been made , the IN All constants in are stored in an array , And this array is ordered . But if there are more values , The consumption is also relatively large . Another example is ：select id from table_name where num in(1,2,3) For continuous values , It works between Don't use in 了 ; Or use connections to replace .

3、 ... and 、SELECT The statement must indicate the field name

SELECT * Add a lot of unnecessary consumption （cpu、io、 Memory 、 network bandwidth ）; Increased the possibility of using overlay indexes ; When the table structure changes , The front end also needs to be updated . So the request is directly in select Followed by the field name .

Four 、 When only one piece of data is needed , Use limit 1

This is to make EXPLAIN in type Column reach const type

5、 ... and 、 If the sort field does not use an index , Just sort as little as possible

6、 ... and 、 If there is no index for other fields in the constraint , Use less as far as possible or

or In the fields on both sides , If there is a non index field , Other conditions are not index fields , Will cause the query not to go through the index . A lot of times union all Or is it union( When necessary ) Instead of “or” Will get better results

7、 ... and 、 As far as possible with union all Instead of union

union and union all The main difference is that the former needs to merge the result set before uniqueness filtering , This involves sorting , A large increase in CPU operation , Increase resource consumption and delay . Of course ,union all The precondition is that there is no duplicate data in the two result sets .

8、 ... and 、 Don't use ORDER BY RAND()

select id from `table_name` order by rand() limit 1000;

Copy

above sql sentence , It can be optimized to

select id from `table_name` t1 join (select  rand() * (select max(id) from `table_name`) as nid) t2 on t1.id > t2.nid limit 1000;

Copy

Nine 、 distinguish in and exists, not in and not exists

select * from  surface A where id in (select id from  surface B)

Copy

above sql Statement equivalent

select * from  surface A where exists(select * from  surface B where  surface B.id= surface A.id)

Copy

distinguish in and exists Mainly caused the change of driving sequence （ This is the key to performance change ）, If it is exists, Then take the outer table as the driving table , Be interviewed first , If it is IN, Then execute the subquery first . therefore IN It is suitable for the case of large exterior and small interior surface ;EXISTS It is suitable for the case of small appearance and large inner surface .

About not in and not exists, Recommended not exists, It's not just about efficiency ,not in There may be a logical problem . How to write an alternative efficiently not exists Of sql sentence ？

primary sql sentence

select colname …  from A surface  where a.id not in (select b.id from B surface )

Copy

efficient sql sentence

select colname …  from A surface  Left join B surface  on where a.id = b.id where b.id is null

Copy

The result set is shown in the figure below ,A The watch is not in B Table data

Ten 、 Use reasonable paging methods to improve the efficiency of paging

select id,name from table_name limit 866613, 20

Copy

Use the above sql When the statement is paginated , Maybe someone will find out , With the increase of table data , Use it directly limit Paging queries will be slower and slower .

The optimization method is as follows ： You can take the maximum number of lines on the previous page id, And then according to this biggest id To limit the starting point of the next page . Than in this column , The biggest one on the previous page id yes 866612.sql It can be written as follows ：

select id,name from table_name where id> 866612 limit 20

Copy

11、 ... and 、 Segmented query

In some user selection pages , Maybe some users choose too large a time range , Causing slow queries . The main reason is that there are too many scan lines . It's time to go through the program , Query in sections , Loop traversal , Combine the results for presentation .

As shown in the figure below sql sentence , Segment query can be used when the number of scanned rows is more than one million

Twelve 、 To avoid the where Field in the clause null Value judgement

about null The engine will give up using the index and scan the whole table .

13、 ... and 、 Not recommended % Prefix fuzzy query

for example LIKE “%name” perhaps LIKE “%name%”, This kind of query will lead to index failure and full table scanning . But you can use LIKE “name%”.

How to query %name%？

As shown in the figure below , Although give to secret Field added index , But in explain Results not used

So how to solve this problem , answer ： Use full text indexing

We often use select id,fnum,fdst from table_name where user_name like '%zhangsan%'; . Such a statement , Ordinary index can't satisfy the query demand . Fortunately, in MySQL in , There is a full-text index to help us .

To create a full-text index sql Grammar is ：

ALTER TABLE `table_name` ADD  FULLTEXT INDEX  `idx_user_name` (`user_name`);

Copy

Using full-text indexing sql The sentence is ：

select id,fnum,fdst from table_name where match(user_name) against('zhangsan' in boolean mode);

Copy

Be careful ： Before you need to create a full-text index , Please contact the DBA Determine if you can create . At the same time, we need to pay attention to the difference between the writing method of the query statement and the common index

fourteen 、 To avoid the where The clause performs an expression operation on the field

such as

select user_id,user_project from table_name where age*2=36;

Copy

In the field on the line of arithmetic operations , This will cause the engine to give up using indexes , It is suggested to change it to

select user_id,user_project from table_name where age=36/2;

Copy

15、 ... and 、 Avoid implicit type conversion

where Clause column Type conversion when the type of the field is inconsistent with the type of the parameter passed in , It is suggested that where Parameter types in

sixteen 、 For federated indexes , Follow the leftmost prefix rule

For example, an index contains fields id,name,school, It can be used directly id Field , It's fine too id,name In this order , however name;school Can't use this index . So when you create a union index, you must pay attention to the order of index fields , Common query fields are placed at the top

seventeen 、 You can use when necessary force index To force the query to go to an index

sometimes MySQL The optimizer takes the index it thinks appropriate to retrieve sql sentence , But maybe the index it uses is not what we want . At this point, it can be used force index To force the optimizer to use the index that we set .

eighteen 、 Note the range query statement

For federated indexes , If there are scope queries , such as between,>,< Under equal conditions , It will invalidate the following index fields .

nineteen 、 About JOIN Optimize

• LEFT JOIN A The table is the driving table
• INNER JOIN MySQL It will automatically find the table with less data to drive the table
• RIGHT JOIN B The table is the driving table

Be careful ：MySQL There is no full join, It can be solved in the following ways

select * from A left join B on B.name = A.name 
where B.name is null
 union all
select * from B;

Copy

Use as much as possible inner join, avoid left join

The tables participating in the union query are at least 2 A watch , In general, there are different sizes . If the connection is inner join, In the absence of other filter conditions MySQL Will automatically select the small table as the driving table , however left join In the selection of driving table, we follow the principle of left driving right , namely left join The table on the left is called the drive table .

Make good use of indexes

The index field of the driven table is used as on Restricted fields for .

Use a small watch to drive a large watch

It can be seen intuitively from the schematic diagram that if the drive table can be reduced , Reduce the number of loops in nested loops , In order to reduce IO Total amount and CPU The number of operations .

Skillfully use STRAIGHT_JOIN

inner join By mysql Select driver table , But in some special cases, you need to choose another table as the driving table , Such as the group by、order by etc. 「Using filesort」、「Using temporary」 when .STRAIGHT_JOIN To force the connection order , stay STRAIGHT_JOIN The table name on the left is the drive table , On the right is the driven table . In the use of STRAIGHT_JOIN There is a precondition that the query is an inner join , That is to say inner join. Other links are not recommended STRAIGHT_JOIN, Otherwise, the query result may be inaccurate .

This approach can sometimes reduce 3 Times of time .

Only the above optimization schemes are listed here , Of course, there are other ways to optimize , You can try to find out , Thank you for attention ..

Wrote last , Welcome to your attention ：http://fenxianglu.cn/