MySQL index

Why not hashmap

1. Fuzzy search cannot be done ,hashcode look for
2. Range lookup

no need B Trees
depth , Reduce IO

b+ The trees are orderly
Left most matching principle

Cluster index ： Primary key ,B+ Trees
Secondary index ：B+ Trees , Leaf nodes store all cluster indexes

Index failure

LOL + - * /
not null

like： Not all , The leftmost matching principle will not fail ,% The beginning will be invalid
or ：or Index fields before and after are not invalidated , If one is not an index, it will become invalid
L（ Joint index ）： Index column order , Start with the first index column

Add, subtract, multiply and divide ：select Field from surface where age -1 = 20; The index will fail
Take the opposite ：not and !=
NULL：is null , is not null , not always , Default values can be set
Use built-in functions ： Function does not work on condition column , Add to = The following values can
Automatic conversion ： Like columns id yes varchar type select XX from surface where id = 123456
123456 yes int type , Will automatically convert the conversion function to varchar. Put a single quotation mark

jion in: Conditions id One is int One is varchar It's going to fail

There is a situation Whether or not to use the index is not necessarily related to the version ,

where a>3

a It's a secondary index
5.6 There were two before
There is a threshold , Less than 20 Ten thousand or less than the average data in the table , Select clustered index
Greater than the secondary index of the selected party
5.6 introduce
3 Time io find id, stay innodb engine Put it in the cache and sort