Asynchronous programming is a form of parallel programming that allows a unit of work to run separately from the primary application thread. When the work is complete, it notifies the main thread (as well as whether the work was completed or failed). There are numerous benefits to using it, such as improved application performance and enhanced responsiveness.
But, like all things in programming, it’s not something that you should use in every instance; in fact, there are some situations in which you should avoid it. Asynchronous programming has actually been around for a long time, but in recent years, it’s become more widely used. According to Mike James at iProgrammer, “Often the programmer is fully aware that what they are doing is object oriented but only vaguely aware that they are writing asynchronous code.” So, we set out to identify the best use cases for async as well as situations in which you shouldn’t use it. We searched the web for insights and reached out to a panel of programmers and asked them to answer this question:
Learn more about the ideal use cases for asynchronous programming by reading what our experts had to say below.
James McCarthyJames is a serial web entrepreneur who builds websites with the aim of helping millions of people. In the last 2 years he has created MRI Directory, Create Brief, and Headphone Charts.
“Every situation is different and there’s no real grab-bag solution…”
As a rule of thumb, asynchronously updating records that are dependent, or depended upon, is generally a bad idea. It’s a quick way to have your data become out of sync. Simple processing of independent data is a great place for asynchronous programming and provides a better UX and, most of the time, better performance in your application.
Calvin BrownCalvin Brown is the Founder of Kairu Consulting.
“Programmers often use async too much…”
In mobile applications to prevent the freezing of the screen, and give the customer the impression that the process is taking actions visually, but it’s usually not. Literally speaking, when an application asks for data (from a database, webservice, etc.), it has to wait until that service replies. All of the fancy gestures that are usually present in applications happen because of async calls to the services at the same time as entertaining the user with some logo or messaging. The amount of resources used at this time is much higher than normal app usage and ultimately, with lots of refreshes, users’ memory suffers.
There are work-arounds, and async is necessary in instances such as preloading data when the app first begins. In other instances, though, it’s cliché and often to the detriment of the users’ device to utilize async so heavily.
Kevin NgKevin Ng is the Tech Lead at Wildebeest.
“Asynchronous programming and parallel programming have great benefits…”
But can be overly used at times. One case I feel programmers may overly use async is for simple and basic computations. There’s no real performance benefits to using async for these calculations.
In addition, another situation when async may not be useful is when you have a single database server not utilizing connection pooling. If all requests hit the same database using a long running connection, it won’t make a difference if the calls are asynchronous or synchronous. Your bottleneck will be the database server.
Lindsey HavensLindsey Havens is the Senior Marketing Manager for PhishLabs.
“Basically you can use Asynchronous programming except when the following conditions are true…”
Asynchronous programming will not provide benefit and actually will result in more overhead on operations that are primarily CPU operations instead of those that involve network or disk overhead.
Manu SinghManu Singh is a Mobile Architect at Clearbridge Mobile with experience in Android Design, Development, and Testing. At Clearbridge, Manu manages project resources for a team of developers building world-class apps for enterprise clients.
“If there is a resource that is used by different elements or a single thread in an application that is responsible for a resource or functionality…”
It needs to be synchronized. For example, on Android, there is a main UI thread, which is responsible for displaying the UI for the app. Having too many asynchronous calls affecting the UI can cause issues and can slow things down. In this case of the UI main thread, you would want some synchronicity so that the UI is displayed properly, there are no conflicts with what is displayed, and the UI population is done in proper sequence.
Steve SilberbergSteve is founder of Fitpacking, a small business that takes people on backpacking adventure vacations to get fit and lose fat.
“If you program Android apps using MIT App Inventor 2…”
You may be forced into asynchronous programming even though you don’t want it. Example: I programmed an app that queries Google Fusion Tables on program load. However, App Inventor 2 continues executing other blocks even though results of the Fusion Table queries have not returned. Because of this, you had better not make these subsequent code blocks dependent upon query results or you will get errors. This is one case where a programmer uses asynchronous programming when they really shouldn’t.
Max GalkaMax Galka is an entrepreneur and web developer based in New York City. He is founder of the mapping platform Blueshift and is an adjunct lecturer at the University of Pennsylvania.
“When transitioning from front-end JavaScript to back-end development with Node…”
The first lesson you learn is to make your loops asynchronous to avoid blocking the event loop. But I often see people take this idea too far. Asynchronous loops are necessary when there is a large number of iterations involved or when the operations within the loop are complex. But for simple tasks like iterating through a small array, there is no reason to overcomplicate things by using a complex recursive function. A simple synchronous for/while loop works just fine, and will also be faster and more readable.
Peter TaskerPeter is a PHP and JavaScript developer from Ottawa, Ontario, Canada. He currently works on the WP Migrate DB Pro team at Delicious Brains. In a previous life he worked for marketing and public relations agencies. He loves WordPress and dislikes FTP.
“Asynchronous code is often used when…”
Writing JavaScript, especially with AJAX handlers or callbacks. It can cause problems when used incorrectly, however. We work with WordPress developers who often run into this when nesting anonymous callback functions. For example, you want to make an AJAX request that depends on an earlier AJAX request. Developers can end up creating nested anonymous callback functions which can lead to messy code and difficulty when debugging errors. With modern JavaScript, this can be avoided with Promises or abstraction to named functions.
Ashish DattaNorth Jersey born and bred, Ashish graduated from Tufts University with a BS in computer science. Following Tufts, Ashish held engineering roles at ArmchairGM and Wikia before founding Setfive Consulting. Outside of the office, you can find Ashish in the gym, out for a run, or sipping cocktails with his pinky out.
“A distinction that many programmers miss is…”
The difference between asynchronous and concurrent programming. At a high level, asynchronous is the ability to do something while waiting for something else to complete, and concurrent is the ability to compute multiple things at the same time, the big trade off vs. sequential execution being that both asynchronous and concurrent programming add additional complexity.
With the explosion in popularity of nodejs, which features an asynchronous environment, programmers are increasingly using asynchronous programming in situations where it offers no benefits. Generally, CPU bound tasks fall into this bucket and examples would be most machine learning algorithms, the transform step in an ETL pipeline, and any sort of cryptocurrency mining.
Johannes KadakJohannes is a Lead Developer at Qminder.
“Synchronous programming is a better fit for…”
Sequential tasks, where stopping the entire program to wait for a network request or disk IO makes sense.
Asynchronous programming is a better fit for code that must respond to events – for example, any kind of graphical UI.
An example of a situation where programmers use async but shouldn’t is any code that can focus entirely on data processing and can accept a “stop-the-world” block while waiting for data to download.
Microsoft Developer NetworkThe Microsoft Developer Network is your resource for development tips, tricks, research, case studies…Everything you need to develop apps that users love.
NOTE: The following information is excerpted from Asynchronous Programming with Async and Await (C# and Visual Basic) via MSDN.
“Asynchrony is essential for activities that are potentially blocking, such as when your application accesses the web…”
Access to a web resource sometimes is slow or delayed. If such an activity is blocked within a synchronous process, the entire application must wait. In an asynchronous process, the application can continue with other work that doesn’t depend on the web resource until the potentially blocking task finishes.
Mark PattersonMark Patterson is a .NET Developer and Founder of HealthSpan Foods.
NOTE: The following information is excerpted from Await Overhead Approximately 50 + (n * 15) ms via Microsoft Developer Network forum.
“Considering the state gymnastics being performed, the Await operator is very efficient. The performance penalty on a fast machine appears to be approximately…”
50 + (n * 15) milliseconds
Where 50 ms is a one time setup delay deferred until runtime the first time an asynchronous method encounters an Await operator. The penalty for each Await operation (including the first one) is approximately 15 ms.
I arrived at these rough estimates by comparing the times required to load images synchronously vs. asynchronously. Obviously your results will vary depending on the speed of your machine, the size of the state being saved and reloaded, and the number of competing threads that might delay scheduling the continuance.
However, this little exploration suggests to me that:
1) It doesn’t make sense to Await tasks which RELIABLY consume <= 20 ms. Obviously an unreliable task, even a very short one – say 1 or 2 ms – could still be a very good candidate for Awaiting. For example, if an IO task fails 1 out of 100 times it should be considered to be very unreliable, and Awaiting the completion of this task would be one way to manage this uncertainty.
2) There is zero penalty for coding an Async method if the Await operator is never encountered. This makes it painless to write methods which can run synchronously or asynchronously depending on an IsAsync parameter passed to the method. In pseudo code something like this:
Private Async Sub DoSomething(ByVal IsAsync as Boolean) Dim Result as T If IsAsync Then Result = Await SomethingTask Else Result = Something End if UseResult(Result) End Sub Private Function Something() as T Dim Result as T ... Return Result End Function Private Function SomethingTask() as Task(of T) Dim f = New Func(of T)(AddressOf Something) Dim s = TaskScheduler.FromCurrentSynchronizationContext Dim tf = New TaskFactory(of T)(s) Return tf.StartNew(f) End Function
(Note the CurrentSynchronizationContext used above is only required if your Async task needs to run on the same thread as the Async method was invoked on. In Silverlight, ANYTHING that touches the UI must be
on the UI thread, for example.)
Brown University Computer ScienceSince 1979, Brown’s Department of Computer Science has forged a path of innovative information technology research and teaching.
NOTE: This information is excerpted from Introduction to Asynchronous Programming via the Brown University Computer Science Program.
“Compared to the synchronous model, the asynchronous model performs best when…”
These conditions almost perfectly characterize a typical busy network server (like a web server) in a
client-server environment. Each task represents one client request with I/O in the form of receiving the
request and sending the reply. A network server implementation is a prime candidate for the asynchronous
model, which is why Twisted and Node.js, among other asynchronous server libraries, have grown so much
in popularity in recent years.
Maria YakimovaMaria Yakimova is a Python/Django developer at Django Stars.
NOTE: The following information is excerpted from Asynchronous Programming in Python | Asyncio (Guide) via Django Stars.
“In a classic sequential programming, all the instructions you send to the interpreter will be executed one by one. It is easy to visualize and predict the output of such a code. But…”
Say you have a script that requests data from 3 different servers. Sometimes, depending on who knows what, the request to one of those servers may unexpectedly take too much time to execute. Imagine that it takes 10 seconds to get data from the second server. While you are waiting, the whole script is actually doing nothing. What if you could write a script that could, instead of waiting for the second request, simply skip it and start executing the third request, then go back to the second one, and proceed from where it left? That’s it. You minimize idle time by switching tasks.
Still, you don’t want to use an asynchronous code when you need a simple script, with little to no I/O.
One more important thing to mention is that all the code is running in a single thread. So if you expect that one part of the program will be executed in the background while your program will be doing something else, this won’t happen.
Brij Bhushan MishraBrij Bhushan Mishra has more than 8 years’ experience and is currently working as an Architect/Consultant. He has had a passion for learning about computers since childhood. He blogs at Code Wala.
NOTE: The following information is excerpted from Concurrency vs Multi-threading vs Asynchronous Programming : Explained via Code Wala.
“There are two things which are very important for any application – Usability and Performance…”
Usability because, say, a user clicks on a button to save some data. This requires multiple smaller tasks like reading and populating data in internal object, establishing a connection with SQL and saving it there, etc. As SQL runs on another machine in the network and runs under a different process, it could be time consuming and may take a bit longer. So, if the application runs on a single thread then the screen will be in a hanged state until all the tasks complete, which is a very bad user experience. That’s why nowadays many applications and new frameworks completely rely on the asynchronous model.
Performance of an application is also very important. It has been seen that while executing a request, around 70-80% of the time gets wasted while waiting for the dependent tasks. So, it can be utilized to the max by asynchronous programming, where once the task is passed to another process (say SQL), the current thread saves the state and is available to take another task. When the SQL task completes, any thread which is free can take it up further.
Eric VogelEric Vogel is a columnist at Visual Studio Magazine. He’s a Senior Software Engineer at Red Cedar Solutions Group and President at Great Lansing User Group for .NET.
NOTE: The following information is excerpted from Asynchronous Programming in .NET: I’ll Call You Back via Visual Studio Magazine.
“Asynchronous programming is a means of parallel programming in which a unit of work runs separately from the main application thread and notifies the calling thread of its completion, failure or progress…”
You may be wondering when you should use asynchronous programming and what are its benefits and problem points.
The main benefits one can gain from using asynchronous programming are improved application performance and responsiveness. One particularly well suited application for the asynchronous pattern is providing a responsive UI in a client application while running a computationally or resource expensive operation.
The .NET Framework provides a few avenues to get on the ramp to asynchronous programming. Some of your implementation choices from the most basic to complex include using background workers, invoking a method asynchronously from a delegate, or implementing the IAsynchResult interface. All of these options allow you to multi-thread your application without ever having to manage your own threads. The .NET Framework asynchronous APIs handle this drudgery for you.
[adinserter block=”33″]
Peter OlsonPeter Olson contributes to hack.guides() on Pluralsight. You can also find him on GitHub.
NOTE: The following information is excerpted from Introduction to asynchronous JavaScript via Pluralsight.
“In asynchronous programs, you can have two lines of code (L1 followed by L2), where L1 schedules some task to be run in the future, but L2 runs before that task completes…”
You can imagine as if you are eating at a sit-down restaurant. Other people order their food. You can also order your food. You don’t have to wait for them to receive their food and finish eating before you order. Similarly, other people don’t have to wait for you to get your food and finish eating before they can order. Everybody will get their food as soon as it is finished cooking.
The sequence in which people receive their food is often correlated with the sequence in which they ordered food, but these sequences do not always have to be identical. For example, if you order a steak, and then I order a glass of water, I will likely receive my order first, since it typically doesn’t take as much time to serve a glass of water as it does to prepare and serve a steak.
Note that asynchronous does not mean the same thing as concurrent or multi-threaded. JavaScript can have asynchronous code, but it is generally single-threaded. This is like a restaurant with a single worker who does all of the waiting and cooking. But if this worker works quickly enough and can switch between tasks efficiently enough, then the restaurant seemingly has multiple workers.
The setTimeout function is probably the simplest way to asynchronously schedule code to run in the future:
// Say "Hello."
console.log("Hello.");
// Say "Goodbye" two seconds from now.
setTimeout(function() {
console.log("Goodbye!");
}, 2000);
// Say "Hello again!"
console.log("Hello again!");
If you are only familiar with synchronous code, you might expect the code above to behave in the following way:
But setTimeout does not pause the execution of the code. It only schedules something to happen in the future, and then immediately continues to the next line.
Joy George KunjikkuruJoy George Kunjikkuru is a passionate software engineer holding deep expertise in the Microsoft technologies stack (.Net Windows Form, ASP.NET, Silverlight, SQL Server) as well as in open web technologies including JavaScript, AngularJS, TypeScript, HTML5, NodeJS, Azure, C#, ASP.Net MVC, WPF, WCF, Linq, and many more. Joy holds a MBA degree from Bharatiar University, Coimbatore and a B.Tech degree in Computer Science from Anna University, Chennai.
NOTE: The following information is excerpted from C# async and await programming model from scratch via Orion.
“Asynchronous programming is essential when we develop any application because it avoids waiting in the main thread on long running operations such as disk I/O, network operations, database access, etc…”
In a normal case, if our program needs something to be done from the results of these long operations, our code is stuck until the operation is done and we proceed from that point.
Using the async mechanism, we can just trigger long running operations and continue with other tasks. These long running operations do the job in a different thread and when they complete it, they notify our main code, and our code can do the post actions from here. When we refer to our code, we are talking about our main thread which deals with the user interface or the thread that primarily processes a web request.
Sathyaish ChakravarthySathyaish Chakravarthy has been programming since early 1997. His area of expertise is the Microsoft .NET framework. He can also work in other languages such as Java, Kotlin, Python, Ruby, JavaScript, and C/C++. He is a freelance full-stack developer. Find him on GitHub and StackOverflow.
NOTE: The following information is excerpted from What is the benefit of using asynchronous programming on a single core machine? via Quora.
“All operations a computer program performs can be categorized into two broad categories…”
Compute-bound operations use the CPU to do some work. Examples are crunching some numbers, doing mathematical calculations such as finding the square root of a number or the value of the constant Pi to a fixed, but rather large number of decimal points, interpolating the missing bits in a bitmap, performing an operation on two bitmaps or two layers of a bitmap, or performing data compression.
I/O-bound operations, on the other hand, do not use the CPU at all. They send a request to the I/O-device-driver and wait for an interrupt from the driver to inform them of the readiness of the result of their requested operation. Examples of I/O-bound operations are reading from a disk or writing to the disk, i.e., reading or writing to files. This can take various forms such as executing a query on a database, reading from or writing to a database, reading from or writing to a network socket, sending a request to a CD-ROM drive, reading or writing to memory, receiving input from the keyboard or mouse, or displaying input on the standard output device, i.e., the monitor, etc.
When a program performs an I/O-bound operation, it sends a request to the underlying operating system API to further direct its request to the appropriate device driver.
Every OS includes both a synchronous version and an asynchronous version of API for I/O-bound operations.
If your computer program makes a call to the synchronous version of your operating system’s I/O-bound API, then while it’s waiting on the results from the device driver, it is essentially blocked from doing anything else. The thread that was allocated to perform the operation simply has to wait for the results. All this time while the thread waits for results from the device driver, it will never be scheduled by the operating system to be on the ready-queue. Consequently, it will get no CPU time. Thus, if your program needs to perform another concurrent operation at this time, it needs an additional thread.
Creating an additional thread involves a significant performance overhead. Also, it reduces scalability since more threads are now doing what needs no doing by any thread, i.e., to simply wait on a device driver.
On the other hand, if your computer program makes a call to the asynchronous version of your operating system’s I/O API, the thread allocated to perform that operation simply issues a request to the device driver and immediately returns. It is now free to do any other work that your program wishes to perform concurrently. If there is nothing required to be done concurrently, it can continue to listen to messages from the operating system’s message queue / dispatcher.
That means if your program is running in an event-driven environment such as a graphical user interface environment, it will provide a more responsive user interface to the user, allowing him to click willy-nilly while the I/O-bound operation is still in progress.
Whether or not your machine has a single core, calling an I/O bound API’s asynchronous version provides better scalability to your application since it obviates the need to create threads when they are not required.
Richard RodgersRichard Rodgers is an entrepreneur, currently Software Architect and Developer at Key and Car Tracking LLC. Previously, he founded TechMate Inc. and has served as the company President since 1993.
NOTE: The following information is excerpted from What situations in programming may we need to use an asynchronous operation? via Quora.
“When you need to wait for another process to complete, it can be very useful to go asynchronous…”
This permits you to continue other operations instead of simply waiting. The operation could be simply a long running process, such as a significant graphic transformation or mathematical calculation. Or, more commonly, it is some type of input/output (IO) operation. Often, you want to run many of these same types of processes simultaneously and asynchronous coding allows you to do that.
Event-driven communications is a common asynchronous paradigm. Basically, you set up communications and provide code, called a callback, that should run when the communication occurs.
For example, I am currently using WebSockets. So, I open the socket and provide callbacks for it including those for receiving a message, recognizing an error, or closing the socket. Whenever a message arrives, my message callback runs. If an error occurs, I handle the error in that callback. If my communication partner terminates, my close callback cleans up my environment. Using these callbacks, I can run off and do something else knowing that, if something happens that I need to handle, the appropriate callback will trigger.
Asynchronous coding often means that you need to multi-thread your code. This means that you have to start another thread that can run independently of your main task. This is often necessary because, as an example, waiting on communication to complete completely stops the thread that is waiting from running. The term for this is “blocking.” If you issue a communication command that is blocking, it will stop and wait for arrival of a message or, if no message arrives, that command will timeout. Running this in a separate thread prevents your main program from having to wait.
Most high-level languages provide libraries that support event-driven and/or asynchronous programming.
Try Stackify’s free code profiler, Prefix, to write better code on your workstation. Prefix works with .NET, Java, PHP, Node.js, Ruby, and Python.
# # #
Asynchronous programming is actually easier than you may think. Check out this post to learn how to return AJAX response from an asynchronous JavaScript call. Or, read this post for more information on how Microsoft has made asynchronous programming simple with the implementation of async await in C# and how the latest versions of ASP.NET are utilizing it to boost performance.
Stackify's APM tools are used by thousands of .NET, Java, PHP, Node.js, Python, & Ruby developers all over the world.
Explore Retrace's product features to learn more.
If you would like to be a guest contributor to the Stackify blog please reach out to [email protected]
