Field notes from production
Practical notes on platforms, reliability, and software delivery that hold up when production gets complicated.
This site collects ideas, cases, and field notes on building and operating systems that teams can change with confidence.
The focus is the DevOps work that shows up outside the diagrams: maintainable automation, useful observability, practical security, developer experience, and delivery decisions under real constraints.
Instead of perfect recipes, this is about operational judgment: tradeoffs, safer paths, and lessons from systems that have to keep running.
Recent posts
-
Before implementing, understand the problem
A solution can work and still solve the wrong problem. Notes on context, reference projects, and engineering judgment before implementation.
-
Not Every Timeout Needs Another Retry
A Kafka metadata timeout in Lambda led us to extend the producer’s recovery window instead of adding another retry loop.
-
Last call before production
How we coordinate multiple deployment lifecycles with previews, releases, and approvals while leaving context-dependent decisions to the people closest to the change.
-
DocumentDB Upserts: When Missing Indexes Drive Up I/O Costs
How missing indexes turned routine DocumentDB upserts into full collection scans, drove I/O cost above $3,000 per month, and slowed every caller down.
-
AWS Infrastructure: Private by Default, Accessible by Design
What private infrastructure means in production: network boundaries, identity, authorization, and intentional access for people and automation.
Labs
Investigate. Decide. Verify.
Practice with simulated incidents and engineering decisions. Inspect the evidence, choose an approach, and verify the outcome.