MySQL 30 military regulations

One 、 The specification （1） You have to use InnoDB Storage engine

Reading ： Support transactions 、 Row-level locks 、 Better concurrency 、CPU And memory cache page optimization for higher resource utilization .

（2） You have to use UTF8 Character set

Reading ： unicode , Without transcoding , No risk of garbled code , Save a space .

（3） Data sheet 、 Data fields must be annotated in Chinese

Reading ：N Who will be in the future tm Know this r1,r2,r3 What are the fields for .

（4） Stored procedures are not allowed 、 View 、 trigger 、Event

Reading ： High concurrency big data Internet business , The architectural design idea is “ Free database CPU, Move the computation to the service layer ”, High concurrency , These features are likely to kill the database , Business logic is more extensible in the services layer , Can be easily implemented “ Add performance to the machine ”. Databases are good at storing and indexing ,CPU Let's move it up .

（5） Do not store large files or photos

Reading ： Why let a database do something it's not good at ？ Large files and photos are stored in the file system , Database storage URI How good .

Two 、 Naming specification

（6） Only Intranet domain names are allowed , instead of ip Connect to database

（7） The online environment 、 development environment 、 Test environment database Intranet domain names follow naming conventions

Business name ：xxx

The online environment ：dj.xxx.db

development environment ：dj.xxx.rdb

Test environment ：dj.xxx.tdb

The slave library is added after the name -s identification , Backup libraries are appended to the name -ss identification

Online from the library ：dj.xxx-s.db

Online storage ：dj.xxx-sss.db

（8） Library name 、 Table name 、 Field name ： A lowercase letter , Underline style , No more than 32 Characters , You must know what you mean , Pinyin is not allowed to be used in English .

（9） Table name t_xxx, Non-unique index name idx_xxx, Unique index name uniq_xxx

3、 ... and 、 Table design specification

（10） The number of singleton tables must be smaller 500

（11） The number of single table columns must be less than 30

（12） The table must have a primary key , For example, self-augment primary keys

Reading ：

a） Increasing the primary key , Row writing can improve insert performance , You can avoid page split , Reduce table fragmentation and memory usage

b） The primary key selects a shorter data type , Innodb The engine's normal index stores the value of the primary key , Shorter data types can effectively reduce index disk space , Improve index caching efficiency

c） Table deletion without primary key , stay row The master-slave architecture of the pattern , Will lead to storage ramming

（13） Foreign keys are not allowed , If there is a foreign key integrity constraint , Application control is required

Reading ： Foreign keys cause coupling between tables ,update And delete All operations involve the associated tables , Very affected sql Performance of , It can even cause a deadlock . High concurrency can easily lead to database performance , The big data high concurrency business scenario database is used to prioritize performance .

Four 、 Field design specification

（14） The field must be defined as NOT NULL And provide default values

Reading ：

a）null The column of / The index statistics / Value comparisons are more complex , Yes MySQL Harder to optimize .

b）null This type of MySQL Special handling is required internally , Increases the complexity of database processing records ; Under the same conditions , When there are more empty fields in the table , The processing performance of the database can be significantly reduced .

c）null Values require more storage space , Either in the table or in each row of the index null Each of the following columns requires additional space for identification .

d） Yes null Processing time , Only use is null or is not null, Can't use =、in、<、<>、!=、not in These operation symbols . Such as ：where name!=’shenjian’, If there is name by null The record of value , The query result is not included name by null The record of value .

（15） No use TEXT、BLOB type

Reading ： It wastes more disk and memory space , A large number of large field queries that are not necessary eliminate hot data , This results in a dramatically reduced memory hit ratio , Impact on database performance .

（16） Do not use decimals to store currency

Reading ： Let's use integers , Decimals can easily lead to misalignment of money .

（17） You have to use varchar(20) Store phone number

Reading ：

a） Refers to the area code or country code , May appear +-()

b） Can phone Numbers do math ？

c）varchar Fuzzy queries can be supported , for example ：like“138%”

（18） No use ENUM, You can use TINYINT Instead of

Reading ：

a） Add new ENUM Value to do DDL operation

b）ENUM The actual internal storage of , You think you're defining a string ？

5、 ... and 、 Index design specification

（19） Single table index is recommended to be controlled in 5 Within a

（20） The number of single index fields is not allowed to exceed 5 individual

Reading ： The field more than 5 Time , Effectively filtering data is no longer possible .

（21） The ban on updates is very frequent 、 Index poorly differentiated attributes

Reading ：

a） Updates change B+ Trees , Indexing frequently updated fields can significantly reduce database performance .

b）“ Gender ” It's a very indistinguishable property , What's the point of indexing , Unable to filter data effectively , The performance is similar to that of a full table scan .

（22） Set up composite index , The highly differentiated fields must be placed first

Reading ： To filter data more efficiently .

6、 ... and 、SQL Use standard

（23） No use SELECT *, Get only the necessary fields , The description column properties need to be displayed

Reading ：

a） Additional columns are read that are not needed CPU、IO、NET Consume .

b） Coverage indexes cannot be effectively utilized .

c） Use SELECT * Easy to appear after adding or deleting fields BUG.

（24） No use INSERT INTO t_xxx VALUES(xxx), The column properties specified for insertion must be displayed

Reading ： Easy to appear after adding or deleting fields BUG.

（25） Property implicit conversion is not allowed

Reading ：SELECT uid FROM t_user WHERE phone=13812345678 Causes a full table scan , And not hit phone Indexes , Guess why ？（ This online problem has come up more than once ）.

（26） Do not use WHERE A function or expression is used on a property of a condition

Reading ：SELECT uid FROM t_user WHERE from_unixtime(day)>='2017-02-15' Causes a full table scan .

The correct way to write it is ：SELECT uid FROM t_user WHERE day>= unix_timestamp('2017-02-15 00:00:00').

（27） Negative queries are not allowed , as well as % The fuzzy query at the beginning

Reading ：

a） Negative query condition ：NOT、!=、<>、!<、!>、NOT IN、NOT LIKE etc. , Causes a full table scan .

b）% The fuzzy query at the beginning , Causes a full table scan .

（28） Do not use large tables JOIN Inquire about , Large tables are not allowed to use subqueries

Reading ： Temporary tables are generated , More memory is consumed CPU, Significantly affects database performance .

（29） No use OR Conditions , Must be changed to IN Inquire about

Reading ： The old version Mysql Of OR The query cannot hit the index , Even if it hits the index , Why make the database more expensive CPU Help with query optimization ？

（30） The application must capture SQL abnormal , And have corresponding processing .