Deep dives on infrastructure, performance, and the decisions behind how Stratum is built.
The work that happens after the response is sent is where reliability quietly breaks. Here is how we made our async tier boring.
A CLI is often the first touch point developers have with your platform, and first impressions compound. Every design decision we made while building the Stratum CLI from scratch — for 80,000 teams across 140 countries — and what we would do differently today.
Networks fail and clients retry. Idempotency keys let a request run exactly once even when it arrives twice. Here is the pattern we use to make writes safe to repeat across the whole platform.
After processing over 10 billion deployment requests, we have learned what it takes to achieve true zero-downtime deploys. The answer involves far more than a simple blue-green traffic switch — it demands coordinated health probing, connection draining, and a rollback system that fires before a single user sees an error.
Logs are the first thing you reach for in an incident and the first bill that spirals out of control. We rebuilt our logging pipeline to stay searchable without breaking the budget.
When a request is slow, the question is always "where?". Tracing turns that guess into a timeline you can read.
Page 4 of 5
POPULAR POSTS
Join 40,000 builders getting our best writing on infrastructure, performance, and reliability. No spam, unsubscribe anytime.
Cloud infrastructure for builders. Deploy globally, scale automatically, sleep soundly.
Product updates and engineering deep-dives, monthly.