A journey to faster feedback on Airflow

Karun Japhet (~karun)


27

Votes

Description:

Over the past four years, our team embraced the challenge of working with Apache Airflow, despite being unfamiliar with it. The absence of clear standards initially resulted in frequent outages.

Within a span of 8 to 10 weeks, we transformed from having no safety nets to never experiencing a system-breaking release again. This was made possible because we automated our test pyramid.

This talk is about the journey that we undertook. It chronicles the process of how we

  1. approached a very unfamiliar domain and technology stack
  2. broke it down into small, understandable pieces
  3. captured that understanding into automated checks
  4. took down our feedback-cum-release cycle from 36 hours down to a mere 90 seconds with nearly 3000 tests.

Prerequisites:

Basic understanding of pytest and the test pyramid

Content URLs:

Slides

Please note

  • Some parts of the slides are a WIP (code samples need to be distilled and improved upon)
  • Slides are subject to change based on sample runs (might add more content and a few more slides to explain concepts better based on feedback from test runs)

Speaker Info:

Karun Japhet is a tech lead, developer, and quality advocate at Sahaj Software. Over the past decade and a half, he has worked on realising value for his clients through the creation of highly scaleable applications and integration of large enterprise applications. More recently, he's been working on building out petabyte scale insights platforms at low cost.

Section: Developer tools and automation
Type: Talks
Target Audience: Intermediate
Last Updated: