Concurrency¶

Concurrency vs Async vs Parallelism

Overview

This document explains the conceptual and practical differences between:

Concurrency
Asynchronous programming (Async)
Parallelism

These terms are often confused, but they describe different aspects of how programs execute tasks.

1. Concurrency¶

Definition¶

Concurrency is a system property.

It means:

The ability of a system to manage multiple tasks during overlapping time periods.

Concurrency does not require tasks to execute at the exact same moment.

Instead, it means:

Multiple tasks can make progress.
The system can switch between tasks.
No task blocks overall progress unnecessarily.

Important Clarification¶

Concurrency is a concept, not an implementation technique.

It can be implemented using:

Multi-threading
Multi-processing
Async/event loops
Actor model
Green threads
Coroutines
Runtime schedulers

Example¶

Even on a single-core CPU:

Task A → executes a bit
Task B → executes a bit
Task C → executes a bit
Switch back to Task A

Tasks appear to run simultaneously, but the CPU is switching between them.

This is concurrency without parallelism.

2. Asynchronous Programming (Async)¶

Definition¶

Async is an implementation strategy for concurrency.

Its main principle:

Do not block a thread while waiting for I/O.

Examples of I/O operations:

Network requests
File reads/writes
Database queries
External API calls

Blocking vs Async¶

Blocking ModelAsync Model

Task A → waits for network
Thread is blocked
CPU may sit idle

Task A → awaits network
Thread becomes free
Task B executes

Async avoids wasting thread time during I/O waits.

Key Idea¶

Async focuses on:

Non-blocking behavior
Efficient handling of I/O-bound workloads
Event-driven scheduling

Important Clarification¶

Async does NOT necessarily mean:

Single-threaded
Parallel
Faster for CPU-heavy work

It only means:

Non-blocking task coordination.

3. Parallelism¶

Definition¶

Parallelism means executing multiple tasks at the exact same time.

This requires:

Multiple CPU cores
Or multiple processors

Example¶

Core 1 → Task A
Core 2 → Task B
Core 3 → Task C

Tasks are physically executing simultaneously.

4. Core Differences¶

Concept	What It Is	Focus	Requires Multiple Cores?
Concurrency	System capability	Managing multiple tasks	No
Async	Implementation strategy	Non-blocking execution	No
Parallelism	Execution model	Simultaneous execution	Yes

5. Relationship Between Them¶

ConcurrencyParallelism

 Concurrency
  ├── Async (non-blocking, event-driven)
  ├── Multi-threaded concurrency
  └── Multi-process concurrency

  Parallelism
    └── True simultaneous execution on multiple cores

6. Deep Conceptual Differences¶

Concurrency is about structure¶

It answers:

Can multiple tasks make progress?

It is about program organization and system capability.

Async is about scheduling behavior¶

It answers:

What happens when a task must wait?

Async says:

Don’t block.
Yield control.
Let something else run.

Parallelism is about hardware execution¶

It answers:

Are tasks literally running at the same time?

7. Critical Distinction¶

A system can be:

Concurrent but NOT Async¶

Multi-threaded blocking example:

Thread 1 → Task A (blocked on I/O)
Thread 2 → Task B
Thread 3 → Task C

This is concurrent. But Thread 1 is still blocked.

Async but NOT Parallel¶

Single-threaded event loop:

Thread 1
  ↳ Task A (await)
  ↳ Task B
  ↳ Task C

This is concurrent and async. But not parallel.

Concurrent AND Parallel¶

Multi-core system:

Core 1 → Task A
Core 2 → Task B
Core 3 → Task C

Tasks run simultaneously and independently.

8. CPU-Bound vs I/O-Bound¶

Idealistic¶

Async is ideal for

I/O-bound workloads
High-latency network operations
Servers handling many connections

Parallelism is ideal for

CPU-bound workloads
Heavy computation
Hashing
Compression
Cryptography
Data processing

Async does not speed up CPU-heavy code.

Parallelism does.

9. Precise Engineering Definitions¶

Concurrency"= the ability to make progress on multiple tasks.
Async = non-blocking coordination of tasks.
Parallelism = simultaneous execution on multiple processors.

10. Common Misconceptions¶

❌ Async means "running at the same time"
✔ No — it means "not blocking."

❌ Concurrency requires multiple cores
✔ No — it can work on a single core.

❌ Async is always faster
✔ Only for I/O-bound workloads.

❌ Parallelism and concurrency are the same
✔ Parallelism is a subset of concurrency.

11. Final Mental Model¶

Think of it this way:

Concurrency → Dealing with many things.
Async → Don’t wait idly.
Parallelism → Doing many things at the same time.

12. Practical Guidance¶

When designing a system:

Use async to handle many I/O operations efficiently.
Use parallelism for heavy CPU computation.
Use concurrency as the architectural principle that allows both.