đŸ‘€ Observability Driven Development

Published 2021-12-27

One of the things I learned quickly while working on large scale systems at Facebook (Meta) was that if your tests passed but you didn't observe your system in production, you had no idea whether or not your system was actually behaving as intended. Without observation in prod, you don't even have a baseline to know what "normal" operation looks like.


o_O Who knows.

All of these things need to be answered to run good software and mantain a high quality bar.

While thinking these problems over and developing best practices for observing systems, I thought I had coined a new term called "Observability Driven Development." After doing some googling, however, I found that a former Facebook Engineering Manager left to become CTO of Honeycomb.io which develops products for exactly this problem and calls it by the same name.

Interesting coincidence, given I'm also coming from Facebook and am planning on releasing tooling that is centered around exactly the same problem as part of my new journey as a MicroFounder. The last coincidental detail was that I had considered reaching out to join the parse team before it was shuttered by FB.

Rather than repeat thoughts explained well elsewhere, here's the article on the same topic by Honeycomb's CTO: A Next Step Beyond TDD