Interview shock 60: what will cause MySQL index invalidation?

author | Lei brother

source | Java Analysis of the real interview questions （ID：aimianshi666）

Please contact authorization for reprint （ WeChat ID：GG_Stone）

In order to verify MySQL Under what circumstances will the index fail , We can use explain Execute the plan to analyze the specific scenario of index failure .

explain Use as follows , Just look up the SQL Add... Before explain Key words can be used , As shown in the figure below ： In the columns of the above query results , Our main observation key This column ,key This column represents the index actually used , If NULL The index is not used , Otherwise, the index is used .

All the results listed above are explained as follows ：

• id — Select the identifier ,id The higher the priority, the higher , The first to be executed ;

• select_type — Represents the type of query ;

• table — Output result set table ;

• partitions — Matching partition ;

• type — Indicates the connection type of the table ;

• possible_keys — When representing a query , Possible indexes ;

• key — Indicates the index actually used ;

• key_len — Length of index field ;

• ref—   Comparison of columns and indexes ;

• rows — Approximate estimated number of lines ;

• filtered — Percentage of rows filtered by table criteria ;

• Extra — Description and description of implementation .

One of the most important is type Field ,type The value types are as follows ：

• all — Scan the full table data ;

• index — Traverse index ;

• range — Index range lookup ;

• index_subquery — Use... In subqueries ref;

• unique_subquery — Use... In subqueries eq_ref;

• ref_or_null — Yes null To optimize the index ref;

• fulltext — Use full text indexing ;

• ref — Find data using a non unique index ;

• eq_ref — stay join Use primary key or unique index association in query ;

• const — Put a primary key in where The following is used as a conditional query , MySQL The optimizer can turn this query optimization into a constant , How and when to transform , It depends on the optimizer , This ratio eq_ref Be more efficient .

  Create test tables and data

  In order to demonstrate and test that situation will lead to index invalidation , Let's create a test table and corresponding data ：

  --  Create table 
  drop table if exists student;
  create table student(
    id int primary key auto_increment comment ' Primary key ',
    sn varchar(32) comment ' Student number ',
    name varchar(250) comment ' full name ',
    age int comment ' Age ',
    sex bit comment ' Gender ',
    address varchar(250) comment ' Home address ',
    key idx_address (address),
    key idx_sn_name_age (sn,name,age)
  )ENGINE=InnoDB DEFAULT CHARSET=utf8;
  --  Add test data 
  insert into student(id,sn,name,age,sex,address) 
    values(1,'cn001',' Zhang San ',18,1,' Gao Laozhuang '),
    (2,'cn002',' Li Si ',20,0,' Huaguoshan '),
    (3,'cn003',' Wang Wu ',50,1,' Waterfall cave ');

  There are... In total in the current table 3 An index , As shown in the figure below ：

  PS： The following content of this article is based on MySQL 5.7 InnoDB Data engine .

Index failure 1： Non leftmost match

The leftmost matching principle refers to , Queries that start with the leftmost field can use a union index , Otherwise, you will not be able to use the union index . The field order of our union index is sn + name + age, We assume that their order is A + B + C, The following union indexes are used as follows ： As can be seen from the above results , If the matching fields start from the leftmost, the upper union index can be used , such as ：

• A+B+C

• A+B

• A+C

  among ：A Equals field sn,B Equals field name,C Equals field age.

and B+C But you can't use the union index , This is the left most matching principle .

Index failure 2： Error fuzzy query

Fuzzy query like The common usage of 3 Kind of ：

1. Fuzzy match followed by any character ：like ' Zhang %'

2. Fuzzy match any preceding character ：like '% Zhang '

3. Fuzzy matching before and after any character ：like '% Zhang %'

And this 3 There are only two kinds of fuzzy queries 1 Three query methods can be used to index , The specific implementation results are as follows ：

Index failure 3： Column operation

If the index column uses an operation , Then the index will fail , As shown in the figure below ：

Index failure 4： Using functions

If any... Is used in the query column MySQL The provided function will invalidate the index , For example, the following columns use ifnull The execution plan after the function is as follows ：

Index failure 5： Type conversion

If the index column has a type conversion , Then I won't go to the index , such as address Is a string type , When querying, you set int A value of type will invalidate the index , As shown in the figure below ：

Index failure 6： Use is not null

When used in a query is not null It can also cause indexes to fail , and is null The index will be triggered normally , As shown in the figure below ：

summary

Lead to MySQL Common scenarios for index invalidation are as follows 6 Kind of ：

1. The union index does not satisfy the leftmost matching principle .

2. The first fuzzy query is an indefinite matching character .

3. The index column is involved in the operation .

4. The index column uses the function .

5. Index column has type conversion .

6. Index columns use is not null Inquire about .

It's up to you to judge right and wrong , Disdain is to listen to people , Gain or loss is more important than number .

official account ：Java Analysis of the real interview questions

Interview collection ：https://gitee.com/mydb/interview

 Previous recommendation 

Java Weekly summary of interview questions （ common 59 piece ）| The latest version

Interview shock 59： Can there be multiple auto increment columns in a table ？

Interview shock 58：truncate、delete and drop Of 6 Big difference ！

Interview shock 57： Cluster index = Primary key index ？