Concurrency Control & NoSQL and new database

concurrency control

1、 Why concurrency control in database ？ What characteristics of transactions can concurrency control technology ensure ？

Share resources when database , Usually multiple transactions are executed at the same time . Simultaneous reads occur when multiple transactions are accessed simultaneously / Write the same data . If you do not control concurrent operations, you may access incorrect data , Destroy the consistency of transactions .

Concurrency control ensures the consistency and isolation of transactions

2、 What kinds of data inconsistencies will be caused by concurrent operations ？ How can we avoid

1） Missing changes , Two transactions modify the same data at the same time , Then the operation of one transaction will be overwritten by the modification of another transaction .

2） It can't be read repeatedly , After a transaction reads some data , Another transaction updates it , Then when you read it again, you will get different data from the last time .

3） Read dirty data , A transaction modifies some data and writes it back to disk , After another transaction reads the data , The previous transaction was cancelled for some reason .

The way to avoid inconsistency is concurrency control , Common concurrency control methods include blocking method 、 Timestamp method 、 Optimistic control 、 Multi version concurrency control method, etc

3、 Blockade and types of blockade

Blockade is business T Before operating on a data , Send a request to the system first , Lock it , After you lock the data, you must have control over the transaction , In the transaction T Before releasing the lock , Other transactions cannot update or read the data

The types of blockade are ： Exclusive lock 、 Shared lock

4、 What problems can the three-level agreement solve

Add X lock , Release until the transaction is over . Solved the problem of missing modifications .

stay 1 Add on the basis S lock , After reading, you can release , Solved the problem of reading dirty data .

stay 1 Add on the basis S lock , Release until the transaction is over , Solved the unrepeatable read problem

5、 Live lock 、 Causes and solutions of live lock

When a series of blocking operations cannot be performed in the correct order , It may cause transactions to wait indefinitely for a blockade .

The way to avoid livelocks is to use FCFS

6、 Deadlock 、 Deadlock solution

Deadlock refers to the execution of two or more processes , A blocking phenomenon caused by competition for resources or communication with each other , If there is no external force , They will not be able to move forward . These processes that are always waiting for each other are called deadlock processes .

There are two ways to prevent deadlock ： Deadlock prevention 、 Deadlock diagnosis and release

There are two ways to prevent deadlocks ： One time blockade 、 Sequential blockade

Deadlock diagnosis and resolution ： Overtime method 、 Transaction wait graph

Relief law ： Choose the transaction with the lowest deadlock cost , Release it

7、 What kind of concurrent scheduling is the correct scheduling

Serializable concurrent scheduling is the correct scheduling . Serializable scheduling definition ： Concurrent execution of multiple transactions is correct , When and

Only if the result is the same as that of serial execution executed in a certain order .

8、 How to ensure the correctness of concurrent scheduling

Conflicts can be serialized , Use two-stage lock protocol

NoSQL and New database

since 1970 Since then , Database technology has developed to the fourth generation , The first generation is 60~70 Hierarchical and mesh database system in early , The second generation is 70 The relational database system started in the mid-s , The third generation is 80 Object database system and 90 Expand the model of , Make the relational database system have object-oriented function .

since 2000 Since then ,Web technology 、 Social networks 、 Advances in mobile computing and the Internet of things have led to semi-structured applications generated worldwide 、 Unstructured data is growing explosively , Such applications create a variety of different requirements , Including horizontal scalability 、 Support large-scale and massive data processing capacity , The system should have high availability and fault tolerance to deal with hardware and software failures , And can respond to client requests by using additional resources , So as to improve the query processing speed , This leads to the fourth generation database technology NoSQL and NewSQL The emergence of Technology .

NoSQL The key feature of a database is ：(1) Use graphics 、 Trees 、 Key value 、 Data models other than tabular models such as documents ;(2) There is no fixed schema definition in database design ;(3) Provide implicit support for horizontal scalability ;(4) Have final consistency rather than strong consistency .NoSQL Another important concept of is CAP Theorem , It provides some constraints on the attributes that are crucial to distributed systems , namely ： Uniformity 、 Availability and partition fault tolerance .CAP The theorem shows that , The distributed system can only meet the following requirements at any given time 2 Attributes .

NewSQL It is the latest generation of database system , It not only provides NoSQL The database system has scalability and scalability

Usability , It also retains the traditional relational database ACID characteristic 、 Relational data model and SQL query language