Free to read ยท Three levels
Apache Kafka: event contracts, processing and dependable recovery
Eighteen developed lessons connect Kafka partition semantics, producer and consumer correctness, transactions, Streams, Connect and operational recovery.
Start learningIn this course
Learn through synthetic parcels, order intents, progress snapshots and bounded offline calculations. Each level contains six lessons with explained examples, failure diagnosis and worked exercises.
The course distinguishes application business identity from log coordinates and keeps every delivery guarantee within its actual system boundary. Complete Arabic content retains the same technical examples and identifiers.
What you will learn
- Design keys, serialized contracts and bounded retention for explicit business requirements.
- Reason through producer retries, consumer progress, duplicate effects and Kafka transaction boundaries.
- Review Streams, Connect, security, capacity and recovery with specific evidence and version assumptions.
Your learning path
Three levels
Basic: records, partitions and reader progress
Model Kafka records, ordering, acknowledgment and retained history.
- Events, topics and the KRaft architecture
- Partitions, keys and ordering
- Producers, batching and acknowledgments
- Consumer groups and polling
- Offsets, commits and replay
- Retention, compaction and tombstones
Intermediate: contracts, failures and observability
Connect schema changes, retries, replication and external effects to diagnostic evidence.
- Serialization and schema evolution
- Idempotence, retries and event identity
- Replication, ISR and write availability
- Poison records and retry topics
- External effects and the outbox pattern
- Lag, throughput and operational evidence
Advanced: atomic processing and recovery design
Review transactional processing, stateful topologies, integration and deployment boundaries.
- Transactions, isolation and fencing
- Streams state, windows and late data
- Connect, converters and change capture
- Listeners, authentication and authorization
- Capacity, partitions and storage budgets
- Recovery, upgrades and cross-cluster replay
Who this is for
- Developers designing retained event flows and replay-safe projections.
- Engineers reviewing messaging incidents, capacity assumptions and recovery evidence.
Before you start
- Basic programming, JSON and application/database concepts; Python is used only for small offline models.
- No broker, connector, database or cloud account is required to study these examples.