Database foundation

---

Database related concepts

• data ： Is the basic object stored in the database , For example, text 、 Images 、 Audio 、 video 、 Students' grades, etc .

• database ： It's stored in a computer for a long time 、 organized 、 Can share A collection of large amounts of data .
  The data in the database is organized according to a certain data model 、 Describe and store .
  have Less redundancy 、 High data independence And scalability , And Share for various users .
  In summary , The database has permanent storage 、 There are three basic characteristics of organization and sharing .

• Database management system （DataBase Management System,DBMS）： It is a data management software between users and operating system , for example MySQL、Redis And so on are database management systems . Its main functions include ： Data definition （ Define the composition and structure of data objects ）; Data organization 、 Storage and management ; Data manipulation （ Add, delete, change, check, etc ）; Database establishment and maintenance ; Transaction management and operation of database （ Secure data 、 Complete, etc ）.

Database classification

According to the different data models used when storing data , There are two main types of databases ： Relational database and Non relational database .

Relational database

Relational database It refers to the use of relational models （ That is, two-dimensional table format ） A database system for organizing data , It mainly includes data lines （ A record ）、 Data columns （ Field ）、 Data sheet （ Set of data rows ）、 database （ Set of data tables ）.
Mainstream relational databases include SQL Server、MySQL、PostgreSQL、SQLite wait .
Advantages of relational database ：

• Easy to understand （ The two-dimensional table structure is very close to the concept of the logical world ）
• Easy to use （SQL Language can easily operate relational databases ）
• Easy to maintain （ The integrity constraint of the relationship reduces the probability of redundant data and data inconsistency ）

Disadvantages of relational database ：

• High concurrency support is insufficient
• Query efficiency is very low in massive data
• It's difficult to scale horizontally （ Adding service nodes is not easy , Downtime maintenance and data migration are often required ）
• Poor performance （ Mainly because of the association query between multiple tables ）

Non relational database

Aiming at the shortcomings of traditional relational database , In order to better adapt to the high concurrency of the Internet 、 High performance 、 The need for high availability and massive data challenges , There is a difference NoSQL（Non-relational SQL） database .NoSQL Give up the relational model and strong transaction guarantee , Focus on the high availability and scalability of the database .
NoSQL Database classification ：

• Key value pair storage ： The mainstream representative is Redis、Memcached
• Document storage ： With relatively free content such as json To store document content , The mainstream representative is MongoDB
• Scalable record storage ： Data is stored in records in , And can accommodate a large number of dynamic columns , The mainstream representative is HBase、Cassandra
• Graph storage ： Store the data in the graph structure , The mainstream representative is Neo4j

Business

Concept

Business Is a user-defined sequence of database operations , All of these operations can be done , Or not at all , It is an indivisible work unit . For example, a SQL A statement can be a transaction .
The beginning and end of a transaction It can be displayed and controlled by the user , If the user does not define transactions explicitly , The database management system will automatically divide the transaction according to the default rules .
Transactions usually end with commit or rollback , Commit will write the updates to the database in the transaction back to disk , Rollback will undo all completed operations on the database in the transaction , Roll back to the state at the beginning of the transaction .

ACID

The transaction has the following 4 A feature ：

• Atomicity （Atomicity）： Transactions are treated as the smallest indivisible unit , All operations of the transaction are either committed successfully , Either all failures roll back .
  Rollback can be implemented by rolling back the log , The rollback log records the modification performed by the transaction , Reverse these operations during rollback .
• Uniformity （Consistency）： The result of transaction execution must be to change the database from one consistency state to another .
• Isolation, （Isolation）： The execution of a transaction cannot be interfered by other transactions , Or the modification made by a firm before submission , Is invisible to other transactions .
• persistence （Durability）： Once the transaction is committed , Its changes will be permanently saved in the database , Even if the system crashes , The result of transaction execution should not be affected .
  When the system crashes, redo logs can be used for recovery , To achieve persistence .

Concurrency consistency

In a concurrent environment , Concurrent execution of multiple transactions , The isolation between transactions is difficult to guarantee , As a result, there are many concurrent consistency problems ：

• Missing changes ： It refers to that the update operation of one transaction overrides the update operation of another transaction . For example, two transactions modify one data at the same time , The first transaction will take effect after being modified , The second transaction is then modified , The modification of the second transaction overrides the modification of the first transaction .
• Dirty reading ： It means that the current transaction reads uncommitted data from another transaction , For example, affairs. T1 Modify a data without submitting , Business T2 Read this data , If T1 The amendment was withdrawn , that T2 The data read is dirty data .
• It can't be read repeatedly ： It refers to reading the same data multiple times in a transaction , But the read data is inconsistent , Because in the meantime , Another transaction modifies this data .
• Fantasy reading ： Illusory reading also belongs to the situation of unrepeatable reading in essence , It happens in T1 Read a range of data ,T2 Insert new data in this range ,T1 When reading the data in this range again , The reading result is different from the first reading .

The main reason of concurrency inconsistency is that the isolation of transaction is destroyed , The solution is ： Ensure isolation through concurrency control .
Concurrency control can be achieved by blocking , But it needs user control , Very complicated .
Database management system provides The isolation level of the transaction , Make it easier for users to deal with concurrency consistency .

The isolation level of the transaction

• Uncommitted read ： Changes in transactions , Even if it's not submitted , It's also visible to other things . Lowest isolation level , Can cause dirty reading 、 It can't be read repeatedly 、 Phantom reading .
• Submit to read ： A transaction can only read the modifications made by the committed transaction . The default isolation level adopted by the large default database , Solved dirty reading .
• Repeatable ： Ensure that the result of reading the same data multiple times in the same transaction is the same . Solve dirty reading and repeatable reading .
• Serializable ： Force transactions to execute serially , Multiple transactions do not interfere with each other , There will be no concurrency consistency issues . The highest isolation level , Solve the above problem 4 A concurrency consistency problem , But this will lead to a decrease in concurrency , Poor performance . It is usually replaced by other concurrency level locking mechanisms .

Reference resources

Database system probability The first 5 edition
https://www.yisu.com/zixun/570175.html
http://www.cyc2018.xyz/%E6%95%B0%E6%8D%AE%E5%BA%93/
If there is any infringement , Please contact the author for deletion