Parallelism, concurrency and several directions for high concurrency optimization

Reprinted , Personal reservation , Forgive me for moving bricks this time , After all, moving bricks is my job ?

1. The difference between concurrency and parallelism

• Concurrent ： When there are multiple threads operating , If there is only one system CPU, hold CPU The running time is divided into several time periods , Assigned to each thread to execute , In a period of time when the thread code runs , Other threads are in suspended state . This way we call concurrency (Concurrent). Concurrent = Interval occurrence
• parallel ： When the system has more than one CPU when , Then the thread operation may be non concurrent . When one CPU When executing a thread , the other one CPU Another thread can be executed , Two threads do not preempt each other CPU resources , You can do it at the same time , This way we call parallel (Parallel).    parallel = At the same time
• High concurrency is one of the factors that must be considered in the design of Internet distributed system architecture , It usually refers to the design to ensure that the system can process many requests in parallel at the same time .
• difference ： Parallel is when two or more events occur at the same time ; Concurrency means that two or more events occur at the same time interval .

2. High concurrency refers to a large number of accesses and requests in a short time , for example 12306 Ticket grabbing and tmall double 11 Activities , Want the system to be able to adapt to high concurrency , It needs to be comprehensively optimized , Include , Hardware 、 The Internet 、 System architecture 、 Selection of development language 、 The use of data structures 、 Algorithm optimization 、 Database optimization …… Multithreading is only one of the solutions .

The role of multithreading in solving the problem of high concurrency is to maximize the utilization of computer resources at every moment , So as not to waste computer resources and make them idle

3. High concurrency needs to be considered ：

• Architecture design of the system , How to reduce unnecessary processing at the architecture level （ Network request , Database operation, etc ）
• Network topology optimization reduces network request time 、 How to design topology , How to implement distributed ？
• Code optimization at the system code level , What design patterns are used to work ？ Which classes need singletons , What needs to be minimized new operation ？
• Improve the efficiency of the code level 、 How to select the appropriate data structure for data access ？ How to design an appropriate algorithm ？
• Synchronous and asynchronous operations at the task execution mode level , Where to use synchronization , Where to use asynchronous ？
• JVM tuning , In order to server Mode or clien mode , How to set up Heap、Stack、Eden Size , How to choose GC Strategy , control Full GC The frequency of ？
• Database optimization reduces query modification time . Database selection ？ The selection of database engine ？ Design of database table structure ？ Database index 、 Trigger design ？ Whether to use read-write separation ？ Still need to consider using data warehouse ？
• Use of cache database , How to choose the cache database ？ yes Redis still Memcache? How to design caching mechanism ？
• Data communication problems , How to choose the mode of communication ？ Use TCP still UDP, Long connection or short connection ？NIO still BIO？netty、mina Still original socket？
• Operating system selection , It's using winserver still Linux？ perhaps Unix？
• hardware configuration ？ yes 8G Memory or 32G, network card 10G still 1G?