Hidden index and descending index in MySQL 8.0 (new feature)

MySQL8.0 Hidden index in

Overview of hidden indexes

Hiding the index means hiding the index , Query in progress 、 When sorting and other operations, the system will directly ignore the index , Hidden index can be understood as soft deletion of index , That is, the index is set to invalid , But in fact, the corresponding index is not deleted , It avoids the accidental deletion of the index , The system reconstruction index needs to consume a lot of time and system resources , Set the hidden index to be visible when necessary , The index returns to its original function

Why should there be hidden indexes ？

In short ： It avoids the accidental deletion of the index , The system reconstruction index needs to consume a lot of time and system resources （ You can deeply experience it through the following two specific situations ）, stay MySQL5.7 You can only rebuild the index , stay MySQL8.0 You can use the hidden index to solve

Scenario analysis 1：
When we delete an index , However, it was found that the performance was seriously affected or some errors occurred after deletion , Then we need to rebuild the index , But in the case of a large amount of data , Rebuilding the index is very time-consuming , So if you can make the index temporarily invalid without deleting the index , Then when an error is found, you don't need to rebuild the index, but restore the index to be valid .
Scenario analysis 2：
When we don't know whether our index is valid , Then delete the index , Whether the index is invalid can be judged by comparing whether there is a difference between the before and after operations . If there is no difference , Then deletion is correct （ Because maintaining indexes requires resources ）, If you find that deleting the index seriously affects the query performance , Then you need to rebuild the deleted index . But delete the index 、 Rebuilding the index requires system resources , So if we can make the index temporarily invalid , Then the above problems can be avoided .

Hide the usage scenario of the index

• When the index needs to be deleted , First, it can be set to hide the index , At this time, the index is invalid , Then you can test , If it is set to hide the index, it has no effect , Then you can use hard deletion , Otherwise, it can be set as non hidden .
• If you want to verify the impact of query performance after an index is deleted , You can temporarily set the index to hidden .

Considerations for hiding indexes

• The primary key cannot be set to hide the index
• When the index is set to hidden, the index is unavailable ( invalid ), But its content will still be updated
• Hidden indexes are invisible to the optimizer , But it can also be visible to the optimizer through parameter settings

1. because InnoDB The storage engine is built with primary keys B+ Trees store data , Therefore, the primary key cannot be set to invalid , Otherwise, the storage engine will reselect a field as the primary key , Rebuild the entire table B+ Trees store data
2. Although the index is set to hidden , But its content is still updated , It's just that indexes won't be used when querying and other operations , So if an index is to be hidden for a long time , Then you'd better delete it , Because it will affect the update of data 、 Efficiency of operations such as deletion

Hidden use code demonstration

• Set hidden indexes when creating tables

CREATE TABLE xiaoyuanhao(
	id INT,
	username INT,
	PASSWORD INT,
	INDEX password_index(PASSWORD) invisible
)
--  By default visible, If you are using invisible Then it's invisible 

• Set the index in the table to hidden

ALTER TABLE xiaoyuanhao ALTER INDEX password_index invisible;

---

MySQL8.0 Descending index in

Overview of descending index

stay MySQL4.0 Although it supports the creation of descending indexes in syntax , however InnoDB The underlying implementation is still the architecture of indexing in ascending order （ Both ascending and descending indexes are built in ascending order ）, In the specific implementation, it is realized by reverse scanning , But this will greatly reduce the efficiency of the database , stay MySQL8.0 Supports the construction of descending index , That is, the underlying index structure is arranged in descending order of index columns

Use and test of descending index

• Create an ascending index , Query in descending order

CREATE TABLE xiaoyuanhao (id INT,PASSWORD INT,username VARCHAR(20));
--  Create a table 
CREATE INDEX username_index ON xiaoyuanhao(username);
--  In ascending order username Create index 
EXPLAIN SELECT username FROM xiaoyuanhao ORDER BY username DESC;
--  Query in descending order 

It can be seen that there is a reverse scan , inefficiency

• Create a descending index , Query in descending order at the same time

DROP INDEX username_index ON xiaoyuanhao;
--  Delete the previous ascending order used 
CREATE INDEX username_index ON xiaoyuanhao(username DESC);
--  Create a descending index 
EXPLAIN SELECT username FROM xiaoyuanhao ORDER BY username DESC;
--  Query in descending order 

It can be seen here that the index is used directly , No reverse scanning , It can be seen that Mysql8.0 The bottom layer of supports the creation of descending indexes , It improves the efficiency of descending query