Pycon.de: serverless orchestration: exploring the future of workflow automation - Tim Bossenmaier¶

Tags: pycon, python

(One of my summaries of the 2025 pycon.de conference in Darmstadt, DE).

What is orchestration? Coordinated execution of multiple commputer systems, applications or services. It is more than automation. Some things you can think of:

• Containers/dockers can be managed.

• Coordinating multiple workflows/tasks.

• Syncronizing/managing two or more apps.

• Coordinating microservices, data services, networks, etc.

You can run code on-prem: a physical server in your cellar or a data center. You can also rent servers from a cloud provider. Another level up is serverless: you pay the specific compute resources you have used. AWS lambda is an example of a serverless function, this popularized the serverless paradigm.

Why would you combine them?

• Resilience: no orchestration tool to keep running.

• Cost efficiency: you only pay for what you use.

• Scalability: automatically handled.

Some options: AWS step functions, azure logic apps, azure durable functions, google’s gcp workflows. A drawback for all of them is that they take a no-code/low-code approach, allowing you to click/drag/drop your workflows in the browser. It is stored in json, so as a proper software developer you are limited to uploading the json with terraform or so.

There are also open source solutions. Argo workflows, for instance. Drawback of those solutions: you again have to worry about infrastructure. If your company already has something set up, it might be a handy idea.

His conclusions:

• Operations: you can orchestrate workflows with minimal OPS overhead.

• Cost: you can build solutions at low cost.

• Perfect for event-driven workflows.

• The future? Probably not (vendor lock-in, mostly). But it is a great extension of your developer-toolbox.

His slides are available on github

Photo explanation: picture from our 2024 vacation around Kassel (DE)