Dig deep into MySQL - resolve the non clustered index of MyISAM storage engine

stay InnoDB Index is data , That is, the cluster index B+ The leaf node of the tree already contains all the complete user records .MyISAM Although the index scheme also uses tree structure , But the index and data are stored separately , This index is also called non clustered index .

create table index_demo(
	c1 int,
	c2 int,
	c3 char(1),
	primary key(c1)
) ROW_FORMAT=COMPACT;

The records in the table are stored separately in a file according to the insertion order of records , This file is not divided into several data pages , As many records as there are, insert as many records into this file , thus , We can quickly access a record through the line number . Use... In the table MyISAM As a storage engine , The representation of its records in the storage space is shown in figure ：

Because there is no deliberate sorting according to the primary key size when inserting data , So we can't use dichotomy to find these data , Use MyISAM The table of the storage engine will store the index information separately in another file , It's called an index file .MyISAM A separate index will be created for the primary key of the table , However, the leaf node of the index does not store complete user records , It's a combination of primary key value and line number . That is, first find the corresponding line number through the index , Then find the corresponding record through the line number .

stay InnoDB In the storage engine , We only need to search the cluster index once according to the primary key value to find the corresponding record ; stay MyISAM In the storage engine , A table return operation is required , It also means that MyISAM The indexes established in are equivalent to all secondary indexes .

MyISAM The address offset of the record in the data file will be stored directly at the index leaf node . From this we can see that MyISAM The operation of returning the table is very fast , Because it takes the address offset directly to the file to get data , and InnoDB It is to find records in the cluster index after obtaining the primary key , Although it's not slow , But it's not as good as directly using an address to access .

If necessary , We can also index other columns separately or jointly , Its principle and InnoDB The index is almost the same , Only the corresponding columns are stored at the leaf node + Line number , These index pages are all secondary indexes .

You can see that for non clustered indexes , Whether the sorting rule is primary key or non primary key , Its structure is the same , That is, the leaf node stores the corresponding columns + Line number .