Java same thread

The same thread is a concurrency model. The meaning of this concurrency model is that a single-threaded system is expanded outward to become N single-threaded systems. The result is N single-threaded systems running in parallel.

A same-threaded system is not a pure single-threaded system because it contains multiple threads. But each of these threads runs like a single-threaded system.

Why is it a single-threaded system?

Today you may be wondering why every system is designed to be single-threaded. Single-threaded systems have gained popularity because their concurrency model is simpler than multi-threading. A single-threaded system does not share any data with other threads. This allows threads to use non-concurrent data structures and better utilize the CPU and CPU cache.

Unfortunately, single-threaded systems will not take full advantage of modern CPUs. A modern CPU often has 2, 4 or more cores. Each core functions as a separate CPU. A single-threaded system utilizes only one of these cores, as shown below:

##Same thread, single-threaded extension

To fully utilize all CPU cores, a single-threaded system can be scaled out to utilize the entire computer.

One thread per CPU

The same thread system usually runs one thread per CPU in the computer. If this computer includes 4 CPUs, or a CPU with 4 cores, then it will standardly run 4 instances of the same thread (4 single-threaded systems), as shown in the diagram:

No shared state

An identical thread system looks similar to a multi-threaded system, because an identical thread has Multiple threads run internally. But there's a subtle difference here.

The difference between the same thread and a multi-threaded system is that there is no shared data in the same thread system. There is no memory accessed by threads at the same time. There are no concurrent data structures and so on. The difference is as shown in the figure below:

The reduction of shared state makes each thread behave as if it is a single-threaded system. However, an identically threaded system can contain more than a single thread, so that it is not truly a "single-threaded system". For lack of a better name, I find it more accurate to call such a system a same-threaded system, rather than a "multi-threaded system with a single-threaded design". The same thread is very simple to say and even simpler to understand.

The same thread mainly means that the data runs in the same thread, and there will be no shared data.

Load Distribution

Obviously, a system of identical threads needs to be shared for this workload between running instances of a single thread. If not, only one instance will get any work, and the system will effectively be single-threaded.

How exactly you distribute load among different instances depends on the design of your system. I'll cover some of that in the sections below.

Single-threaded microservices

If your system contains multiple microservices, each microservice can run in a single thread mode. When you deploy multiple single-threaded microservices to the same computer, each microservice can run a separate thread on a separate CPU.

Microservices do not share any data, so microservices are a good use case for a same-threaded system.

Services with sharded data

If your system does not need to share data, or at least a database, you may be able to split it piece of this database. Sharding means that the data is divided into multiple databases. The data is representatively separated so that related data are located in the same database. For example, all data belonging to some "owner" entity will be inserted into the same database. Sharding is outside the scope of this tutorial, so you will have to check out some tutorials on the subject.

Thread communication

If threads in the same thread need to communicate, they do this via message passing. A thread wants to send a message to thread A, it can do this by generating a message (a sequence of bytes). Thread B can then copy the entire message (sequence of bytes) and read it. By copying the message, thread B determines that when it modifies the message, thread A cannot modify the message. Once it is copied, it is immutable to thread A.

The schematic diagram of this process is as follows:

#This thread communication can be replaced by queues, pipes, Unix sockets, and TCP sockets. Whatever works for your system.

Simpler concurrency model

Every system running in its own thread can use the same thread system if it is single-threaded To achieve. This means that the internal concurrency model becomes simpler than threads with shared state. You don't need to worry about concurrent data structures and all concurrency issues.

Illustration

Here is an illustration of a single-threaded, multi-threaded, same-threaded system. You can get an overview of the differences more easily

The first diagram shows a single-threaded system:

The second image shows a multi-threaded system with shared data:

The third image shows an identical thread with two threads with separated data systems communicate with each other by passing information.

The above is the content of the same Java thread. For more related content, please pay attention to the PHP Chinese website (www.php.cn)!