Streams

Unix philosophy, MapReduce, join algorithms, Hadoop vs MPP, Spark/Flink dataflow engines, Pregel graph processing

Observing DB changes as event stream; log-based (Debezium, WAL/binlog/oplog), trigger-based, query-based mechanisms; CDC vs event sourcing; dual-write problem solved by outbox + CDC; data liberation patterns; strangler migration; schema coupling; log-compacted topics for bootstrap; backpressure on consumers

CDC, event sourcing, CQRS, immutability, write/read path, correctness without coordination; DDD: CQRS sync/async projections, architectural slices, event-sourced domain model four-step cycle; Fowler's four-way event taxonomy (notification/state-transfer/event-sourcing/CQRS)

Log-based brokers (Kafka), stateless topology primitives, partition assignment, windowing, late event strategies, watermarks vs stream time, reprocessing, state stores (internal/external, changelog, hot replicas), effectively once processing, heavyweight vs lightweight framework comparison, stream joins, exactly-once fault tolerance; Kafka production mechanics: producer batching (acks/idempotence), consumer commit semantics, semantic partitioning, consumer groups, ISR