Writing

Prometheus, Grafana, and Loki are easy to wire up. Deciding what deserves a page at 2am is the actual hard part.
Read article
TLS protects data in transit, but it doesn't stop a payload from being read the moment it lands somewhere you didn't expect — logs, proxies, or a compromised intermediary.

Connection pooling isn't a problem you feel until you have too many services opening connections at once — I wired it in early instead of waiting for the outage.

The instinct when a system slows down is to fix it fast. What actually worked was slowing down first — to understand the problem before touching anything.