Blog
Para quem já sabe o básico e quer ir fundo. Aqui o assunto é como os modelos funcionam em produção: memória, roteamento, ferramentas, agentes. O lado técnico que pouca gente explica direito.
Brokers, controller (KRaft 2024+), topics, partitions, replication factor, ISR, leader election. Producer idempotence. Consumer group rebalance. Tiered storage.
Confluent Schema Registry. Backward/forward/full compatibility. Avro vs Protobuf vs JSON Schema trade-offs. Schema evolution sem quebrar consumers.
Change Data Capture via binlog (MySQL, Postgres WAL, Mongo oplog). Debezium connectors. Snapshot vs streaming. Handling schema changes. Dual writes são bugs.
At-most, at-least, exactly-once (EOS). Producer idempotence + transactions. Consumer read-process-write patterns. Limitations (external side effects).
DB write + Kafka publish atomicamente. Outbox table + CDC ou transactional outbox. Idempotent consumers. Por que dual writes falham em 10% das vezes.
Stream processing stateful. KStreams, KTables, windows, joins. ksqlDB SQL sobre streams. Quando Streams vs Flink vs Spark Streaming.
Entregáveis: pipeline (OLTP Postgres → Debezium → Kafka → Flink/Streams → OLAP ClickHouse), schema registry, outbox pattern, exactly-once. Benchmark throughput.