[MySQL learning notes 23] index optimization

Check the execution frequency of the database

show global status like ‘com_______’;

Slow query log

Introduce

The slow query log records all events that have exceeded the specified time （ Default 10s） Of SQL sentence .
Default mysql Do not enable slow query

To configure

my.cnf
# Turn on mysql Slow log switch
slow_query_log=1
# Set the time of slow log to 2s
long_query_time=2

Check the log

Open file localhost-slow.log You can see it.

profile

Introduce

profile yes mysql Functions provided to monitor the execution of query statements

View your mysql Does the version support profile function

select @@have_profiling;

If it shows YES, It means that

Turn on profile

set [session|global] profiling = 1;
# If specified session Then it will only take effect for the current session , If specified global It is effective for the global 

see profiles

show profiles;# Query each item sql Time consuming 
show profile for query query_id;# View specified id Of sql Statement details take time 
show profile cpu for query query_id;# View specified id Of sql sentence cpu usage 

explain Implementation plan

Usage method

stay sql Add a... Before the statement explain perhaps desc Key words can be used

explain select * from emp where id=5;

Execution results

Partial column meaning

id: Primary key
select_type: Express select type , Common values are simple( A simple watch , That is, it does not apply to table joins or subqueries ),primary（ Main query , That is, the outer query ）,union（union The second or subsequent statement in ）,subquery（ Subquery ）
type: Indicates the type of access , Indicates the connection type , The order of performance from good to bad is NULL、system、const、eq_ref、ref、range、index、all
possible_key: Indexes that may be referenced
key: Actual index
key_len: Number of bytes of index used
rows: Number of rows to execute the query , It's a valuation
filtered: Indicates the percentage of returned rows in the number of rows to be read ,filtered The greater the value, the better

Index optimization

The leftmost prefix rule

If you index multiple columns , Follow the leftmost prefix rule . The leftmost prefix rule means that the query starts from the leftmost part of the index , And do not skip the columns in the index . If you skip a column , Then this column and all subsequent indexes will be invalidated .

Range queries

If a range query appears in the union index , Column index failure on the right side of the range

Column operation

If the index column is involved in the operation , Then the index is invalid

String of characters

If you don't use quotation marks for character strings , You don't use indexes . Without single quotation marks, there will be implicit type conversion or direct error reporting , Don't go by index .

Fuzzy query

If the tail is fuzzy , The index works , If the header performs a fuzzy query , Index does not work .

or Inquire about

Use or When inquiring , As long as one of the conditions has no index , Then the whole will fail

Data distribution prediction

If mysql It is predicted that the efficiency of using indexes is not as good as that of full table scanning , The whole table will be scanned directly , For example, there is a field that is age,age The minimum value of is 10, So if one condition is age>=10,mysql Will directly scan the whole disk , because age>=10 You can see that the data volume covers the whole table , There's no need to use an index .

SQL Tips

use index

by sql Statement provides a reference index
select * from table use index(name,age);

ignore index

tell mysql Do not use certain indexes
select * from table ignore index(name,age);

force index

mandatory mysql Use the specified index
select * from table_name force index(index_name) where xxx;

Overlay index

The columns of the query should be included in the index as much as possible , This is efficient , Otherwise, you need to query back to the table to get the value of the target column

Prefix index

If some fields are long strings , Using this field to create an index will consume a lot of space , Do more harm than good ,mysql Provides us with a prefix index , You can extract the first few characters of a string as an index , Reduce index space .
create index index_name on table_name(column(n));

Selection of single column index and joint index

Imagine this situation
info Table structure
id： Primary key
name： full name
phone： Telephone
And all three fields have indexes ,id Corresponding to the unique index （ Aggregate index ）,name and phone It's ordinary. B+Tree Indexes
When we subquery like this ：
select id,name,phone from info where name=‘xxx’ and phone=‘xxxx’;
because name and phone They're all single column indexes ,mysql Will choose one of them to use , Not both .
Suppose the choice is name Indexes , So according to name The only data found in the index is name and id Information , did not phone Information , Therefore, you must perform a back table query to find out the complete data .
And when we work for name,phone After building a union index ,mysql We only need to query the joint index once to return the information we need , One back to table query is missing , More efficient .