Concurrency

Concurrency is the composition of independently executing tasks so that a program can deal with many things at once. It is about structure: organizing a program into tasks that can run in overlapping time windows. Crucially, concurrency does not require true simultaneous execution — a single CPU core can be concurrent by rapidly switching between tasks.

How It Works

Concurrent programs use constructs such as threads, coroutines, async/await, actors, or event loops to manage multiple in-flight tasks. When one task waits — for disk, network, or a lock — the runtime can switch to another that is ready, keeping resources busy. Because tasks may access shared state, concurrency introduces hazards like race conditions, deadlocks, and data corruption. Languages and libraries address these with locks, channels, atomic operations, and immutable data. The safest concurrent designs minimize shared mutable state.

Why It Matters

Modern software is full of waiting: requests to databases, calls to remote services, user input. Concurrency lets a program stay responsive and throughput-efficient by overlapping that waiting with useful work. A web server handling thousands of simultaneous connections relies on concurrency to avoid blocking on each one. Concurrency also makes it possible to exploit parallel hardware, though the two ideas are distinct.

Getting concurrency right is famously hard; subtle bugs only appear under specific timing. This is why immutability and pure functions, which remove shared mutable state, are valued in concurrent code.

Related Terms

Parallelism is about executing tasks simultaneously, often confused with concurrency. Immutability and pure functions make concurrent code safer. Idempotent operations help when tasks repeat.