Chapter 11: Streaming Processing

Stream = never-ending sequence of events (records).
Each event is an immutable fact that happened at a moment in time (e.g., “user 123 bought item A at 14:03 UTC”).
Contrast with a file in batch jobs (fixed, bounded).

Term	Plain meaning	Analogy
Producer (publisher)	Code that emits events	Web server writing access log lines
Consumer (subscriber)	Code that reads events	Fraud detector that reacts to payments
Topic / stream	Named category of events	“purchases”, “temperature-sensor-42”
Event encoding	How data is stored on the wire (JSON, Avro, …)	File format for each record

Producers might out-run consumers. Systems must pick ONE strategy:

TCP (Transmission Control Protocol) pipes, UDP (User Datagram Protocol) multicast, webhooks, ZeroMQ.
Very low latency, but little durability: if the consumer is offline, events vanish or require custom retransmit logic.