blog.log
// cloud, DevOps, incident response, and things I only learned the hard way
featured.md
A walkthrough of the audit, the low-risk wins we shipped first, and the two changes that actually moved the needle — right-sizing instances wasn't one of them.
Read the full postarchive.json
Three years of infrastructure-as-code across a dozen client projects, distilled into the module patterns worth keeping — and a graveyard of the ones that weren't.
A cascading failure, a retry storm, and the one-line config change that would have prevented three hours of downtime — with the postmortem template I use now.
A honest framework for deciding between Kubernetes, plain containers, and managed PaaS — based on team size, not what's trending on Hacker News.
A pragmatic rollout plan for zero-trust — starting with identity-aware proxies instead of ripping out your VPN on day one.
Blameless doesn't mean toothless. The postmortem structure I use with clients to get honest timelines and action items that actually get done.
The strangler-fig approach to containerizing a decade-old monolith, service by service, while it stayed in production the whole time.