Apache Kafka
Apache Kafka от основ до прода: топики и партиции, группы потребителей, гарантии доставки, хранение, Spring Kafka, очередь разбора, реестр схем и отставание.
Why it matters for UCP. Kafka is the channel for event-driven architecture in UCP services: domain events between bounded contexts, CDC from the database, analytics. At maturity level 2+ Kafka is almost always already in the stack. Part of the preparation program.
Two levels of depth. Fundamentals cover how Kafka itself works and what guarantees it gives. The production guide covers what is specific to Java/Spring and what you need to know to operate it.
Section articles
- Fundamentals: topics, partitions, guarantees — what is inside Kafka, how consumer groups work, which delivery guarantees exist, idempotent / transactional producer, replication, retention. Language-agnostic.
- Kafka in production: Spring Kafka, DLQ, Schema Registry, tuning — Spring Kafka (
@KafkaListener,KafkaTemplate), error handling and DLQ, Schema Registry + Avro/Protobuf, consumer lag, performance tuning (batch.size/linger.ms/compression), security (SASL/SSL/ACL), KRaft instead of ZooKeeper.
Related
- Distributed patterns — Saga, Outbox, Idempotent Consumer: how Kafka is used in real systems.
- Kafka Style Guide — the rules for working with Kafka in our Java/Spring services (codes
R-KAFKA-*). - Resilience patterns — Circuit Breaker, Timeout, Retry: applicable to Kafka consumers.