B+ tree (5) introduction to MyISAM -- MySQL from getting started to mastering (17)

In the last article, we said , A federated index uses more than two columns to create an index ,b+ The tree is a , First use c2 Column sorting , If the result is the same , Then use c3 Sort .innoDB Of b+ The characteristic of tree is that the root node remains unchanged , The new table has a clustered index by default , First, there is a root node without data , Put the user record data into the root points , When the data is slow , Page splitting , There will be many nodes , At this moment, the root node evolves into a root directory record node , Data is stored in the underlying node . The inner nodes of the secondary index ensure consistency , Add a primary key when saving column values .

B+ Trees (4) Joint index --mysql From entry to mastery （ sixteen )

myISAM brief introduction

We know innoDB The index of search engine is data , It is divided into list value index tree , And clustered index tree , Cluster index that b+ A tree index is data , All user records exist in leaf nodes . and myISAM Although it is also B+ Trees , But indexes and data are separate .

myISAM Yes, the user record data is all put in one file , It's called a data file , This file will not be divided into several pages , Just stuff as many records as you have , Through the corresponding line number of each line, you can quickly access , Because it is inserted at will , It will not be sorted by primary key size , Cannot use binary search method to find .

Then the index information will be stored in another index file ,myISAM The index will be created separately for the primary key of the table , Only the user record data is not stored in the leaf node , It's the primary key + Line number , It means that after finding the corresponding line number , I will use the line number to find all the data in the data file again . therefore myISAM Every query must be returned to the table , Equivalent to secondary index .(innoDB The cluster index of is to find the corresponding inner node directly in the root directory record page according to the primary key , Find all the data on the corresponding underlying leaf node ).

If necessary , You can also create secondary indexes or joint indexes , Only the leaf node stores column values + Line number .

Create index and delete index ：

So much theoretical knowledge , How to fight in practice ？mysql Medium innoDB and myISAM The table will automatically be the primary key or declared as unique To create a clustered index , But if you need to create secondary indexes for other columns , You need to in sql Show indication in . Why not create it automatically , Don't forget that more indexes mean more b+ Trees , It means that both insert and delete require huge memory to maintain b+ Trees , It's very performance intensive .

When creating tables, you can create indexes ,index and key Keywords can be used , Appoint indx_tb_col Tabular c2 As a idx_c2 Index of names ：

create table index_tb_col(
  c1 int,
  c2 int,
  c3 char(1),
  index idx_c2 (c2)
)row_format=Compact;

Copy

Or after the table is created , Appoint c3 by idx_c3 Index of names ：

mysql> alter table index_tb_col add index idx_c3 (c3);
Query OK, 0 rows affected (0.04 sec)

Copy

You can also add federated indexes and delete indexes ：

// Composite index 
mysql> create table index_tb_col(
    -> c1 int,
    -> c2 int,
    -> c3 char(1),
    -> primary key (c1),
    -> index idx_c2_c3 (c2,c3)
    -> )row_format=Compact;
Query OK, 0 rows affected (0.03 sec)
 
// Delete index 
mysql> alter table index_tb_col drop index idx_c2_c3;
Query OK, 0 rows affected (0.03 sec)

Copy