Abstract

”1202 ALARM! 1202 ALARM!” These words from Buzz Aldrin during Apollo 11’s lunar descent weren’t a crisis announcement - they were proof that Margaret Hamilton’s resilient software design worked exactly as intended. When faced with an overflow of data, the Apollo Guidance Computer prioritized critical tasks and kept going, ultimately landing humans on the moon with just seconds of fuel remaining.

Today’s data pipelines face similar challenges: complex systems, distributed dependencies, and high-stakes outcomes where failure is inevitable. As data workflows evolve from monolithic ETL jobs into distributed microservices, they demand the same resilience patterns that revolutionized application development. This talk explores how the lessons from Apollo - from Hamilton’s priority scheduling to NASA’s rigorous engineering mindset - directly inform modern data engineering practices and are embodied in Prefect 3.0’s transactional orchestration.

We’ll examine:

• How Apollo’s approach to failure management mirrors modern chaos engineering principles
• Why Hamilton’s “engineering mindset” is more relevant than ever in the age of AI
• How Prefect 3.0’s transactional approach implements these historical lessons
• Practical patterns for building resilient data pipelines that gracefully handle failure across distributed services

Whether you’re building ETL pipelines, training ML models, or managing the “long tail” of operational data workflows, these lessons from the space age are more relevant than ever.

Don’t panic: we will explore together how the spirit of Apollo can help us build more resilient data systems for the future.