An interactive study companion for Martin Kleppmann's classic on reliable, scalable, and maintainable systems.
Event streams deliver records continuously — via message brokers, partitioned logs, or direct sockets.
Unlike request-response, event streams deliver a sequence of records over time. Message brokers decouple producers and consumers. Partitioned logs add durability and replay — consumers can re-read history at their own pace.
Apache Kafka is the default partitioned log for microservices — order events, audit logs, and CDC streams. RabbitMQ and AWS SQS route task queues with at-least-once delivery. For browser real-time updates, WebSockets (bidirectional) and Server-Sent Events (server push over HTTP) complement the async backend. Kafka transactions plus idempotent producers achieve exactly-once processing semantics within a cluster.
Message delivery events append to a partitioned log so multiple consumer groups — push notifications, analytics, moderation — read the same stream at independent offsets without competing for messages.
// LinkedIn / Uber-style Kafka consumer with offset tracking
async function consumeMessages(groupId: string) {
const consumer = kafka.consumer({ groupId });
await consumer.subscribe({ topic: "user-events" });
await consumer.run({
eachMessage: async ({ partition, message }) => {
const event = JSON.parse(message.value!.toString());
await processEvent(event); // must be idempotent for at-least-once
// offset committed after processing (or in transaction with side effect)
},
});
}