Fault Tree Analysis Applied to Apache Flink

on YouTube ↗35m

" summary

Applies classic Fault Tree Analysis to Apache Flink, models Kafka, ZooKeeper, Kubernetes, S3 and checkpoints, and ships an SLO calculator plus step‑by‑step formulas. Shows counterintuitive result that adding task managers can lower availability, quantifies an example: 20 task managers, 5‑minute checkpoints, 60‑minute poison‑pill MTTR, monthly patches yields 99.44% within 20 minutes (≈4 hours error budget/month). Emphasizes checkpoint mean+sigma tuning, local recovery, and tooling to estimate tradeoffs.

AI-written summary · about

" chapters (6)

1. 0:00Introduction
2. 2:08Critical Use Cases
3. 5:27Introduction to Fault Tree Analysis
4. 10:58Flink Fault Trees
5. 24:56SLO Calculator
6. 30:40Summary