A Redis Job Queue Was Enough
We almost reached for Kafka. Then we counted our jobs per second and put BullMQ on Redis instead. Two years later, here is why we have not regretted it.
Engineering Journal
Notes from the platform
Tradeoffs, failures, and design decisions from building an AI-infrastructure learning platform. Postgres, Redis, real-time systems, vector databases, Kubernetes, AIOps — and what actually breaks in production.
Browse by topic
Recent posts
All posts →
/ Backend Decisions
Write to Postgres Before You Publish the Event
The order in which you commit a row and publish a real-time event determines whether your UI and your database can ever disagree. Always write first.
/ Backend Decisions
Why PgBouncer in Transaction Mode Is the Default
Postgres tops out at a few hundred concurrent connections. Workers and serverless handlers want thousands. Transaction-mode pooling is the cheap fix — and a small set of constraints you have to live with.
/ From Failure Detection to Autonomous Remediation
Detection Is Easy. Remediation Is the Hard Part.
Why most AIOps tools stop at alerting, and what it takes to build systems that actually fix problems.
/ AIOps Fundamentals
Anomaly Detection in Monitoring: A Deep Dive
Explore different anomaly detection techniques and how to choose the right approach for your infrastructure monitoring needs.
/ Dependency & Network Failure Intelligence
Why Network Semantics Matter More Than Packet Loss
Understanding semantic network failures like DNS issues, certificate expiry, and configuration errors that cause real outages.