Java Multithreading Tutorial
Threads are the backbone of multithreading. We are living in a real-world which in itself is caught on the web surrounded by lots of applications. With the advancement in technologies we cannot achieve the speed required to run them simultaneously unless we introduce the concept of multi tasking efficiently. It is achieved by the concept of thread.
Suppose you are using two tasks at a time on the computer, be it using Microsoft Word and listening to music. These two tasks are called processes. So you start typing in Word and at the same time start music app, this is called multitasking. Now you committed a mistake in a Word and spell check shows exception, this means Word is a process that is broken down into sub-processes. Now if a machine is dual-core then one process or task is been handled by one core and music is been handled by another core.
In the above example, we come across both multiprocessing and multithreading. These are somehow indirectly used to achieve multitasking. In this way the mechanism of dividing the tasks is called multithreading in which every process or task is called by a thread where a thread is responsible for when to execute, when to stop and how long to be in a waiting state. Hence, a thread is the smallest unit of processing whereas multitasking is a process of executing multiple tasks at a time.
Multitasking is being achieved in two ways:
- Multiprocessing: Process-based multitasking is a heavyweight process and occupies different address spaces in memory. Hence, while switching from one process to another, it will require some time be it very small, causing a lag because of switching. This happens as registers will be loaded in memory maps and the list will be updated.
- Multithreading: Thread-based multitasking is a lightweight process and occupies the same address space. Hence, while switching cost of communication will be very less.
Below is the Lifecycle of a Thread been illustrated
- New: When a thread is just created.
- Runnable: When a start() method is called over thread processed by the thread scheduler.
- Case A: Can be a running thread
- Case B: Can not be a running thread
- Running: When it hits case 1 means the scheduler has selected it to be run the thread from runnable state to run state.
- Blocked: When it hits case 2 meaning the scheduler has selected not to allow a thread to change state from runnable to run.
- Terminated: When the run() method exists or stop() method is called over a thread.
If we do incorporate threads in operating systems one can perceive that the process scheduling algorithms in operating systems are strongly deep-down working on the same concept incorporating thread in Gantt charts. A few of the most popular are listed below which wraps up all of them and are used practically in software development.
- First In First Out
- Last In First Out
- Round Robin Scheduling
Now Imagine the concept of Deadlock in operating systems with threads – how the switching is getting computed over internally if one only has an overview of them.
So far we have understood multithreading and thread conceptually, so we can conclude advantages of multithreading before moving to any other concept or getting to programs in multithreading.
- The user is not blocked as threads are independent even if there is an issue with one thread then only the corresponding process will be stopped rest all the operations will be computed successfully.
- Saves time as too many operations are carried over at the same time causing work to get finished as if threads are not used the only one process will be handled by the processor.
- Threads are independent though sharing the same address space.
So we have touched all main concepts of multithreading but the question striving in the head is left. why do we need it, where to use it and how? Now, we will discuss all three scenarios why multithreading is needed and where it is implemented via the help of programs in which we will be further learning more about threads and their methods. We need multithreading in four scenarios as listed.
- Thread Class
- Mobile applications
- Asynchronous thread
- Web applications
- Game Development
Note: By default we only have one main thread which is responsible for main thread exception as you have encountered even without having any prior knowledge of multithreading
Two Ways to Implement Multithreading
Method 1: Using Thread Class
Java provides Thread class to achieve programming invoking threads thereby some major methods of thread class are shown below in the tabular format with which we deal frequently along the action performed by them.
|isDaemon()||It checks whether the current thread is daemon or not|
|start()||It starts the execution of the thread|
|run()||It does the executable operations statements in the body of this method over a thread|
|sleep()||It is a static method that puts the thread to sleep for a certain time been passed as an argument to it|
|wait()||It sets the thread back in waiting state.|
|notify()||It gives out a notification to one thread that is in waiting state|
|notifyAll()||It gives out a notification to all the thread in the waiting state|
|setDaemon()||It set the current thread as Daemon thread|
|stop()||It is used to stop the execution of the thread|
|resume()||It is used to resume the suspended thread.|
Pre-requisites: Basic syntax and methods to deal with threads
Now let us come up with how to set the name of the thread. By default, threads are named thread-0, thread-1, and so on. But there is also a method that is often used as setName() method. Also corresponding to it there is a method getName() which returns the name of the thread be it default or settled already by using setName() method. The syntax is as follows:
(a) Returning the name of the thread
public String getName() ;
(b) Changing the name of the thread
public void setName(String name);
Taking a step further, let us dive into the implementation part to understand more concepts about multithreading. So, there are basically two ways of implementing multithreading:
Illustration: Consider if one has to multiply all elements by 2 and there are 500 elements in an array.
Thread1 is running Thread2 is running
Here we have created our two thread classes for each thread. In the main method, we are simply creating objects of these thread classes where objects are now threads. So in main, we call thread using start() method over both the threads. Now start() method starts the thread and lookup for their run() method to run. Here both of our thread classes were having run() methods, so both threads are put to run state from runnable by the scheduler, and output on the console is justified.
Thread 1 is running
Here we have created our two thread classes for each thread. In the main method, we are simply creating objects of these thread classes where objects are now threads. So in main, we call thread using start() method over both the threads. Now start() method starts the thread and lookup their run() method to run. Here only class 1 is having the run() method to make the thread transcend from runnable to run state to execute whereas thread 2 is only created but not put to run state by the scheduler as its corresponding run() method was missing. Hence, only thread 1 is called rest thread 2 is created only and is in the runnable state later blocked by scheduler because the corresponding run() method was missing.
Thread 2 Thread 1 is running
Method 2: Using Runnable Interface
Another way to achieve multithreading in java is via the Runnable interface. Here as we have seen in the above example in way 1 where Thread class is extended. Here Runnable interface being a functional interface has its own run() method. Here classes are implemented to the Runnable interface. Later on, in the main() method, Runnable reference is created for the classes that are implemented in order to make bondage with Thread class to run our own corresponding run() methods. Further, while creating an object of Thread class we will pass these references in Thread class as its constructor allows only one runnable object, which is passed as a parameter while creating Thread class object in a main() method. Now lastly just like what we did in Thread class, start() method is invoked over the runnable object who are now already linked with Thread class objects, so the execution begins for our run() methods in case of Runnable interface. It is shown in the program below as follows:
Thread2 Thread1 Thread2 Thread1 Thread2 Thread1 Thread2 Thread1 Thread2 Thread1
Points to remember: Whenever you wanted to create threads, there are only two ways:
- Extending the class
- Implementing the interface which is runnable
Make sure to create an object of threads in which you have to pass the object of runnable
Special Methods of Threads
Now let us discuss various methods that are there for threads. Here we will be discussing major methods in order to have a practical understanding of threads and multithreading which are sequential as follows:
- start() Method
- suspend() Method
- stop() Method
- wait() Method
- notify() Method
- notifyAll() Method
- sleep() Method
- Output Without sleep() Method
- Output with sleep() method in Serial Execution Processes (Blocking methods approach)
- Output with sleep() method in Parallel Execution Processes (Unblocking methods approach)
- join() Method
Note: For naive users in multithreading where threads are backbone go through Program 4 to get very basics of threads, how to start, make it hold, or terminate then only toggle to program 1 and rest as follows.
Shot Shot Shot Shot Shot Miss Miss Miss Miss Miss
Case 2: Video output
Shot Shot Shot Shot Shot Miss Miss Miss Miss Miss
Case 3: Video output
Shot Miss Shot Miss Shot Miss Shot Miss Shot Miss
Note: There is no priority been set for threads for which as per the order of execution of threads outputs will vary so do remember this drawback of multithreading of different outputs leading to data inconsistency issues which we will be discussing in-depth in the later part under synchronization in threads.
Priorities in Threads
Priorities in threads is a concept where each thread is having a priority which is represented by numbers ranging from 1 to 10.
- The default priority is set to 5 as excepted.
- Minimum priority is set to 0.
- Maximum priority is set to 10.
Here 3 constants are defined in it namely as follows:
- public static int NORM_PRIORITY
- public static int MIN_PRIORITY
- public static int MAX_PRIORITY
Let us discuss it with an example to get how internally the work is getting executed. Here we will be using the knowledge gathered above as follows:
- We will use currentThread() method to get the name of the current thread. User can also use setName() method if he/she wants to make names of thread as per choice for understanding purposes.
- getName() method will be used to get the name of the thread.
Case 1: Default Priority
Running Thread : Thread-0 Running Thread : Thread-1 Running Thread Priority : 5 Running Thread Priority : 5
Case 2: NORM_PRIORITY
Running Thread : Thread-0 Running Thread : Thread-1 Running Thread Priority : 5 Running Thread Priority : 5
Case 3: MIN_PRIORITY
Running Thread : Thread-0 Running Thread : Thread-1 Running Thread Priority : 1 Running Thread Priority : 1
Case 4: MAX_PRIORITY
Running Thread : Thread-1 Running Thread : Thread-0 Running Thread Priority : 10 Running Thread Priority : 10
If we look carefully we do see the outputs for cases 1 and 2 are equivalent. This signifies that when the user is not even aware of the priority threads still NORM_PRIORITY is showcasing the same result as to what default priority is. It is because the default priority of running thread as soon as the corresponding start() method is called is executed as per setting priorities for all the thread to 5 which is equivalent to the priority of NORM case. This is because both the outputs are equivalent to each other. While in case 3 priority is set to a minimum on a scale of 1 to 10 so do the same in case 4 where priority is assigned to 10 on the same scale.
Hence, all the outputs in terms of priorities are justified. Now let us move ahead onto an important aspect of priority threading been incorporated in daily life – Daemon thread
Daemon thread is basically a service provider thread that provides services to the user thread. The scope for this thread start() or be it terminate() is completely dependent on the user’s thread as it supports in the backend for user threads being getting run. As soon as the user thread is terminated daemon thread is also terminated at the same time as being the service provider thread.
Hence, the characteristics of the Daemon thread are as follows:
- It is only the service provider thread not responsible for interpretation in user threads.
- So, it is a low-priority thread.
- It is a dependent thread as it has no existence on its own.
- JVM terminates the thread as soon as user threads are terminated and come back into play as the user’s thread starts.
- Yes, you guess the most popular example is garbage collector in java. Some other examples do include ‘finalizer’.
Exceptions: IllegalArgumentException as return type while setting a Daemon thread is boolean so do apply carefully.
Note: To get rid of the exception users thread should only start after setting it to daemon thread. The other way of starting prior setting it to daemon will not work as it will pop-out IllegalArgumentException
As discussed above in the Thread class two most widely used method is as follows:
|Daemon Thread Methods||Action Performed|
|isDaemon()||It checks whether the current thread is a daemon thread or not|
|setDaemon()||It set the thread to be marked as daemon thread|
Let us discuss the implementation of the Daemon thread before jumping onto the garbage collector.
Another way to achieve the same is through Thread Group in which as the name suggests multiple threads are treated as a single object and later on all the operations are carried on over this object itself aiding in providing a substitute for the Thread Pool.
While implementing ThreadGroup do note that ThreadGroup is a part of ‘java.lang.ThreadGroup’ class not a part of Thread class in java so do peek out constructors and methods of ThreadGroup class before moving ahead keeping a check over deprecated methods in his class so as not to face any ambiguity further.
Here main() method in itself is a thread because of which you do see Exception in main() while running the program because of which system.main thread exception is thrown sometimes while execution of the program.
It is the mechanism that bounds the access of multiple threads to share a common resource hence is suggested to be useful where only one thread at a time is granted the access to run over.
It is implemented in the program by using ‘synchronized‘ keyword.
Now let’s finally discuss some advantages and disadvantages of synchronization before implementing the same. For more depth in synchronization, one can also learn object level lock and class level lock and do notice the differences between two to get a fair understanding of the same before implementing the same.
Why synchronization is required?
- Data inconsistency issues are the primary issue where multiple threads are accessing the common memory which sometimes results in faults in order to avoid that a thread is overlooked by another thread if it fails out.
- Data integrity
- To work with a common shared resource which is very essential in the real world such as in banking systems.
Note: Do not go for synchronized keyword unless it is most needed, remember this as there is no priority setup for threads, so if the main thread runs before or after other thread the output of the program would be different.
The biggest advantage of synchronization is the increase in idiotic resistance as one can not choose arbitrarily an object to lock on as a result string literal can not be locked or be the content. Hence, these bad practices are not possible to perform on synchronized method block.
As we have seen humongous advantages and get to know how important it is but there comes disadvantage with it.
Disadvantage: Performance issues will arise as during the execution of one thread all the other threads are put to a blocking state and do note they are not in waiting state. This causes a performance drop if the time taken for one thread is too long.
As perceived from the image in which we are getting that count variable being shared resource is updating randomly. It is because of multithreading for which this concept becomes a necessity.
- Case 1: If ‘main thread’ executes first then count will be incremented followed by a ‘thread T’ in synchronization
- Case 2: If ‘thread T‘ executes first then count will not increment followed by ‘main thread‘ in synchronization
Implementation: Let us take a sample program to observe this 0 1 count conflict
Here the count is incremented to 1 meaning ‘main thread‘ has executed prior to ‘created thread‘. We have run it many times and compiled and run once again wherein all cases here main thread is executing faster than created thread but do remember output may vary. Our created thread can execute prior to ‘main thread‘ leading to ‘Count : 0’ as an output on the console.
Now another topic that arises in dealing with synchronization in threads is Thread safety in java synchronization is the new concept that arises out in synchronization so let us discuss it considering
- A real-life scenario followed by
- Pictorial representation as an illustration followed by
- Technically description and implementation
Suppose a person is withdrawing some amount of money from the bank and at the same time the ATM card registered with the same account number is carrying on withdrawal operation by some other user. Now suppose if withdrawing some amount of money from net banking makes funds in account lesser than the amount which needed to be withdrawal or the other way. This makes the bank unsafe as more funds are debited from the account than was actually present in the account making the bank very unsafe and is not seen in daily life. So what banks do is that they only let one transaction at a time. Once it is over then another is permitted.
Interpreting the same technology as there are two different processes going on which object in case of parallel execution is over headed by threads. Now possessing such traits over threads such that they should look after for before execution or in simpler words making them synchronized. This mechanism is referred to as Thread Safe with the use of the keyword ‘synchronized‘ before the common shared method/function to be performed parallel.
As we know Java has a feature, Multithreading, which is a process of running multiple threads simultaneously. When multiple threads are working on the same data, and the value of our data is changing, that scenario is not thread-safe, and we will get inconsistent results. When a thread is already working on an object and preventing another thread from working on the same object, this process is called Thread-Safety. Now there are several ways to achieve thread-safety in our program namely as follows:
Conclusion: Hence, if we are accessing one thread at a time then we can say thread-safe program and if multiple threads are getting accessed then the program is said to be thread-unsafe that is one resource at a time can not be shared by multiple threads at a time.
- Java Program to illustrate Incomplete Thread Iterations returning counter value to Zero irrespective of iteration bound
- Java Program to Illustrate Complete Thread Iterations illustrating join() Method
- Java Program to Illustrate thread-unsafe or non-synchronizing programs as of incomplete iterations
- Java Program to Illustrate Thread Safe And synchronized Programs as of Complete iterations using ‘synchronized‘ Keyword.
Count : 0
Count : 10000
Count : 151138
Count : 200000
In case 1 we can see that count is zero as initialized. Now we have two threads main thread and the thread t1. So there are two threads so now what happens sometimes instance is shared among both of the threads.
In case 1 both are accessing the count variable where we are directly trying to access thread via thread t1.count which will throw out 0 always as we need to call it with the help of object to perform the execution.
Now we have understood the working of synchronization is a thread that is nothing but referred to as a term Concurrency in java which in layman language is executing multiple tasks. Let us depict concurrency in threads with the help of a pictorial illustration.
Consider the task of multiplying an array of elements by a multiplier of 2. Now if we start multiplying every element randomly wise it will take a serious amount of time as every time the element will be searched over and computer. By far we have studied multithreading above in which we have concluded to a single line that thread is the backbone of multithreading. So incorporating threads in the above situation as the machine is quad-core we here take 4 threads for every core where we divide the above computing sample set to (1/4)th resulting out in 4x faster computing. If in the above scenario it had taken 4 seconds then now it will take 1 second only. This mechanism of parallel running threads in order to achieve faster and lag-free computations is known as concurrency.
Note: Go for multithreading always for concurrent execution and if not using this concept go for sequential execution despite having bigger chunks of code as safety to our code is the primary issue.