Foundations & the Log
Kafka as a distributed log, the append-only abstraction, brokers/topics/partitions/offsets, and KRaft. Start →
Foundations & the Log
Kafka as a distributed log, the append-only abstraction, brokers/topics/partitions/offsets, and KRaft. Start →
Topics, Partitions & Replication
Partitioning and ordering, keys, replication and the ISR, retention and log compaction. Learn →
Producing Messages
The producer API, acks and durability, the idempotent producer, batching and compression. Produce →
Consuming & Consumer Groups
The poll loop, consumer groups and rebalancing (KIP-848), offset commits, and delivery semantics. Consume →
Delivery Semantics & Transactions
End-to-end guarantees, transactions (KIP-890), exactly-once processing, and error handling. Guarantee →
Streams, Connect & Schema
Schema Registry, Kafka Connect, Kafka Streams, and choosing a processing layer. Process →
Operations & Production
Running Kafka locally, monitoring and consumer lag, tuning, and security. Operate →
The model first, then the code
Every topic starts from a model — the log, the partition, the consumer group, the transaction — so Kafka’s behavior becomes mechanism you understand, not configuration you guess at.
Kafka 4.x, verified
Written against Kafka 4.x and checked against the official docs via context7 — KRaft (no ZooKeeper), the idempotent producer, the KIP-848 consumer rebalance protocol, and KIP-890 transactions are covered as they exist today.
Guarantees, not just APIs
Not “here’s send()” but “here’s what acks=all plus min.insync.replicas actually guarantees, and where a message can still be lost or duplicated.” You’ll reason about delivery with confidence.
Diagrams that explain
The log, partition assignment, replication and the ISR, and the transactional loop as theme-aware Mermaid diagrams. Bilingual (English / ไทย) with saved quiz progress.