Microservices, Observability, SLOs, Releases & Disaster Recovery
Question bank: service boundaries, telemetry that works during a disaster, percentile maths, SLI/SLO/SLA contracts, error-budget policy and burn-rate alerts, canaries, feature flags, Multi-AZ, multi-region and DR drills. New angles only, no repeats of the deep-dive topics.
About 95 minutes · 30 exam questions · 21 interview questions · 2 code labs
What you will learn
- Microservices — boundaries, costs, and when to merge back
- Observability — context, cardinality, and surviving the disaster
- P95/P99 — percentiles, tail latency, and aggregation pitfalls
- SLA / SLO / SLI — measure the user, promise less than you target
- Error budgets — policy first, then burn-rate alerts
- Canary deployment — a statistical experiment, not a waiting room
- Feature flags — decouple deploy from release, then clean up
- Multi-AZ — surviving a datacentre, not a region
- Multi-region — latency, residency, and hidden single-region dependencies
- Disaster Recovery — RPO, RTO, corruption, and proving it