deployment practice · Self-paced
Kafka Clients for Java Services
Producer reliability, consumer group rebalances, and idempotent pipelines with plain Java clients.
- Duration
- 4 weeks · 44 mentor hours
- Schedule
- Flexible
- Level
- Intermediate
- Price
- 12,400 THB (informational)
Kafka 3.7 · Java 21 · Micrometer
Description
You implement idempotent producers, exactly-once-ish processing patterns, and lag dashboards using Micrometer + Prometheus scrape configs. Labs assume a single-broker local stack plus mocked schema registry.
What arrives in your repo
- Producer acks + linger tuning workbook
- Consumer partition assignment tracing
- Dead-letter topic design patterns
- Schema compatibility dry runs (JSON + Avro)
- Replay harness with time-bound offsets
- Chaos exercises for broker bounce
- TH latency assumptions baked into SLAs
Outcomes we expect you to evidence
- Stand up a reproducible local topology
- Author runbook for rebalance storms
- Define lag alert thresholds with evidence
Responsible instructor
Somchai Meesuk
Streaming lead for loyalty platforms.
FAQ
Mentioned in appendix only; client APIs remain the focus.
No — local Docker only to keep costs predictable.
See Money-Back Policy — 14 days from purchase if no lab submitted.
Experience notes
“Kafka Clients for Java Services replay harness caught our offset commit ordering bug pre-prod.”