Updated: Jun 3, 2022
Singleton Design Pattern is a creational design pattern. Singleton pattern lets you ensure that only one instance of a class (struct in go) is created. This single instance created is called singleton object or singleton instance. Singleton pattern lets you ensure that a class has only one instance, thereby providing a global access to this instance.
The singleton pattern usually provides a global method which is responsible for creating and returning this singleton instance. This global method wraps the object creation logic thereby hiding the complexity from the client. The user of this object might not even realize that he is working with a singleton object.
When we say singleton object is accessible globally, it is important we make sure that we wont allow clients to modify this global object. Once the singleton object is created the pattern should ensure that it is immutable. Otherwise the client or application may end up seeing an unexpected behaviour while using this object.
The singleton pattern ensures that the object is created only once, and once it is created the pattern returns the same object without any modification, no matter how many times the client tries to get the object. To ensure that the singleton object is accessed globally and that it cannot be modified once created, singleton pattern encapsulates the singleton object inside a globally accessible method, also you need to make the instance unexported which guarantees it is not accessible directly outside the package.
Use cases where singleton pattern is useful:
Database connection object: In case of a database, creating a database connection object is usually an expensive operation since it needs a lot of heavy-lifting to be done in the background. So, ideally what we would want in this case is to create the database connection object once and reuse it across our application rather than creating a new object for each call. This makes singleton pattern an ideal choice for this use case.
Logger: Even in case of logging for an application, enabling or disabling logs, changing the severity of logged messages etc needs to be done on a single instance which makes singleton pattern suitable for this case.
As mentioned earlier, we need to provide a global GetInstance() method to our users. The first caller who calls this method creates the singleton instance. For rest of the callers we return the same instance which was created earlier.
As simple as it may seem, it is very easy to get the singleton pattern wrong. Lets discuss some of the correct ways to create the singleton pattern in go.
In go the singleton pattern can be implemented in two ways mainly:
Using mutex locks.
Using the Once.Do() method in sync package.
Using Mutex Locks
Below is a implementation using mutex locks:
The idea here is simple. During the program execution, the first thread or go routine to call the GetInstance() method acquires the lock and continues execution to the next line of code. After this point no matter how many threads try to call the GetInstance() method, they will be locked at line 13 in code, they wont be allowed to proceed further since the first thread has already acquired the lock.
The first thread satisfies "if instance == nil" condition and therefore enters the if block and creates a singleton instance (line 17) and returns it (line 19).
Once the first thread returns from GetInstance() method the mutex lock is released and next thread (in order as they are called) acquires the lock. But now onward "if instance == nil" condition is always false since the first thread has already created a singleton instance. Therefore all threads after the first thread simply return from the GetInstance() method without creating a new instance. Thus the GetInstance() method returns a single instance of singleton struct no matter how many times it is called.
This approach works fine and does serve its purpose of creating a singleton instance. But there is one slight drawback in the above implementation. Mutex locks can be expensive because it needs to do some bookkeeping in the background to ensure that no more than one thread or go routine acquires lock at a time, among other things. This is especially true when there are multiple threads trying to acquire a lock. In the above implementation, if there are 100 calls to GetInstance() method, all 100 threads will acquire a lock before returning the singleton instance. But we know acquiring a lock after the first thread (after the singleton instance is created) is unnecessary. The purpose of using mutex lock is to ensure that if multiple threads try to access the GetInstance() method at the same time, in parallel, the GetInstance() methods doesn't create multiple instances (multiple copies) of our singleton struct.
However, we can modify the above implementation slightly to overcome this drawback. All that we have to do is wrap lines 13-18 within another "instance == nil" check. This way all calls to GetInstance() method that come after the first call doesn't have to acquire a lock, our GetInstance() method can simply return the singleton instance after the outer "instance == nil" check.
Below is the modified solution:
The first (outer) if statement checks if the thread needs to acquire lock, and the second (inner) if statement tell us if the singleton instance is already created or we need to create it.
Want to master coding? Looking to learn new skills and crack interviews? We recommend you to explore these tailor made courses:
Using "Once.Do()" Method
Another way to implement the singleton pattern in golang is by using the Once.Do() method in sync package. The library ensures that the code within once.Do() block is executed only once, this guarantee is implicitly provided by go.
Note: To know more about the internal working of Once.Do() method refer this.
Below is the implementation using Once.Do():
In this method we place the singleton instance creation code inside the Once.Do() block instead of using the mutex locks. The Once.Do() method also uses mutex locks internally, so it is necessary to use the outer if statement even for this approach.
We can test both these implementations using the the below code snippet:
The above snippet creates 1000 go routines which call the GetInstance() method almost in parallel. You can notice from the below output that both of our solutions print the "Creating a new singleton instance" message only once indicating only one instance of the singleton struct is created even though 1000 goroutines are invoking the GetInstance() method.
That is all for this article, thank you for taking your time to read this. If you have any questions or doubts, please let us know in the comments section below, we will be happy to answer you.
If you found this article useful, do not forget to subscribe to our website, your support motivates us to bring out more such articles in future (scroll down to the bottom of the page to find the subscription form).
You can explore more such amazing articles from code recipe in our blogs section.
Get 100% discount on Code Recipe Membership Plan. Join now and get exclusive access to premium content for free. Hurry! Offer only available for a limited time. Join now.