How would you choose and operate Uvicorn or Gunicorn workers for a Python API?

Treats Python web workers as measured processes with separate memory, lifecycle, and failure boundaries.

1Say itAnswer out loud first

2CompareCompare with Core answer

3RefineRefine with Stronger and Advanced

TL;DR

Uvicorn is an ASGI server and can run one or multiple worker processes. Gunicorn is a process manager with worker classes; an ASGI application needs an ASGI-capable worker arrangement. Each worker is a separate process with its own event loop, connections, and application state. I start with the deployment platform's process model, then choose worker count from CPU, memory, blocking behavior, and measured throughput rather than a fixed formula.

Full answer below

Practice focus

Treats Python web workers as measured processes with separate memory, lifecycle, and failure boundaries.

Answer first

Answer path

Core answer first, then checkpoints toward a stronger answer.

Start
Core answer
Start here. This is the direct spoken answer to practice first.
Overview
Treats Python web workers as measured processes with separate memory, lifecycle, and failure boundaries.
Uvicorn is an ASGI server and can run one or multiple worker processes. Gunicorn is a process manager with worker classes; an ASGI application needs an ASGI-capable worker arrangement. Each worker is a separate process with its own event loop, connections, and application state. I start with the deployment platform's process model, then choose worker count from CPU, memory, blocking behavior, and measured throughput rather than a fixed formula.
Checkpoint
Stronger answer
Optional upgrade with examples, trade-offs, and sharper phrasing.
More workers can use more CPU cores and isolate failures, but they multiply connection pools and much of the application's memory. In an orchestrator, one worker per container with more replicas can make health, rollout, and resource limits simpler; on a single host, a manager with several workers may be appropriate. Reload mode is for development, not a production replication strategy. The supervisor or platform must restart failed processes.

Review signals

Use these to catch weak phrasing, shallow reasoning, and risky claims.

Common mistakesCandidateCommon mistakesMistakes that weaken the answer or signal shallow understanding.Open details

Details

Copying a CPU-times-two formula blindly
Ignoring per-worker pools and memory
Using development reload in production

How would you choose and operate Uvicorn or Gunicorn workers for a Python API?

Practice focus

Answer path

Overview

Review signals

Follow-up prompts

Continue practice

Next interviewer turn

Tech stacks

Skill areas

Suggest an improvement

Found this answer useful?