We turned off Pub/Sub and nobody noticed
Summary
Incident.io details migrating from a single Google Cloud Pub/Sub broker to a dual-broker setup (NATS and Pub/Sub), using an eventadapter abstraction, dynamic load balancing, and chaos testing to achieve higher availability and observability. The article highlights design decisions like oldest-message-first scheduling and cross-broker failover with circuit breakers, reporting tangible reliability improvements.