Skip to content

Fan-Out / Fan-In — Junior

At junior level, focus on this question:

Why does splitting one task across multiple independent workers give you real parallelism, and what makes a task splittable this way?

Fanning out: independent sub-tasks run concurrently

import concurrent.futures

def process_chunk(chunk):
    return sum(x * x for x in chunk)

chunks = [data[i:i+1000] for i in range(0, len(data), 1000)]

with concurrent.futures.ThreadPoolExecutor(max_workers=4) as executor:
    results = list(executor.map(process_chunk, chunks))  # FAN-OUT
flowchart LR Data["10,000 items"] --> Split["Split into 10 chunks\nof 1,000 each"] Split --> W1[Worker: chunk 1] Split --> W2[Worker: chunk 2] Split --> WN[... Worker: chunk 10] Note["Each chunk is INDEPENDENT -\nno worker needs to know\nabout another's progress"]

Fan-out works specifically because each chunk's processing is independent — worker 2 doesn't need any information from worker 1's work to do its own. This is the exact "embarrassingly parallel" data- parallelism shape referenced throughout the parallel-programming topics in this same folder.

🎓 Takeaway: fan-out gives you real parallelism precisely when the work naturally decomposes into independent pieces — if sub-tasks genuinely depend on each other's results, you don't have a fan-out opportunity at all, you have a sequential dependency chain.

Test yourself

  1. Why does independence between chunks matter for fan-out to actually provide parallelism benefit?
  2. Give an example of a task that CANNOT be naively fanned out because its sub-parts depend on each other.
  3. If you fan out 10,000 items across 4 workers, roughly how many items would each worker handle, and why does uneven chunk sizes matter for overall completion time?

Continue to middle.md.