Async/Await Complexity and Its Impact on Concurrency in Production

Overview of Async/Await

In the past decade, async/await has become popular due to its syntax that resembles synchronous programming. This familiarity allows developers to write asynchronous code with relative ease, making it an appealing choice for software development. However, this ease of use comes with significant underlying complexity.

Complexity in Operation

Rich Hickey highlighted this distinction between ease and simplicity in programming, suggesting that while async/await is easy to implement, it conceals structural complexities that burden the developer. Rob Pike echoed this sentiment at GopherConAU, noting that while async/await is simpler for language designers compared to alternatives like goroutines, it shifts operational complexity back to programmers.

The Trap of Async/Await

The primary issue arises from the way async/await conflates asynchrony with concurrency. Developers may employ async functions as if they were straightforward, sequential code, neglecting the implications of CPU-bound tasks. When extensive computations are performed in an async context, they can stall the entire execution thread, leading to increased latency for multiple simultaneous requests.

Performance Implications in Production

In cooperative runtimes like Tokio and Node.js, tasks do not yield control back to the event loop until they encounter an await point. If a CPU-bound task executes, it can severely impact the performance of unrelated asynchronous operations, risking system responsiveness and underutilizing computing resources. This highlights the necessity for developers to think critically about how they structure their asynchronous functions.

Conclusion: Rethinking Concurrency Models

The article emphasizes the importance of properly separating runtimes and managing the complexities of concurrent programming. Understanding the limitations of async/await can help developers create more efficient applications and avoid the pitfalls associated with its use in production environments.