Nomad Job Scheduling with Python — ELI5

Imagine you run a bakery with five ovens. Each morning, you have a list of things to bake — bread, cakes, cookies. You need to figure out which oven has space, which one is the right temperature, and how to keep everything running even if one oven breaks down.

Nomad is like a smart baker who handles all that scheduling for your Python programs across many computers.

When your Python app needs to run somewhere — on a server in the cloud, inside a Docker container, or even as a plain script — Nomad decides where it runs, starts it, and watches it. If the program crashes, Nomad restarts it. If a server goes down, Nomad moves the program to another server.

This is called job scheduling. You describe what you want to run (“my FastAPI app, 3 copies, each needing 512MB of memory”), and Nomad figures out the rest. You don’t choose which server — Nomad picks the best one based on available resources.

Python developers interact with Nomad through its HTTP API. Python libraries like python-nomad let you submit jobs, check their status, and scale them up or down — all from Python code. This means your deployment scripts, monitoring tools, and automation pipelines can control Nomad directly.

HashiCorp (the company behind Terraform and Vault) built Nomad as a simpler alternative to Kubernetes. While Kubernetes is powerful, it’s complex. Nomad trades some features for simplicity — making it popular with teams that want container orchestration without the steep learning curve.

The one thing to remember: Nomad is a job scheduler that automatically places and manages your Python programs across multiple servers, and Python can control it all through a simple API.