Why should you care?

Millions of audit events per day need replay, retention, and parallel consumers — RabbitMQ queues differ semantically.

See it live — copy this example

Paste into your MeanVerse project (Angular + Express + MongoDB), then run with ng serve / node / mongosh as noted.

import { Kafka } from 'kafkajs';

const kafka = new Kafka({ brokers: [process.env.KAFKA_BROKER!] });
const producer = kafka.producer();
await producer.connect();
await producer.send({
  topic: 'meanverse.audit',
  messages: [{ key: tenantId, value: JSON.stringify({ action: 'TRANSFER', userId, at: Date.now() }) }]
});

const consumer = kafka.consumer({ groupId: 'analytics-writer' });
await consumer.subscribe({ topic: 'meanverse.audit' });
await consumer.run({ eachMessage: async ({ message }) => {
  await AuditWarehouse.insert(JSON.parse(message.value!.toString()));
}});

What happened?

• Topic retains events by retention policy.
• key tenantId keeps tenant events in same partition order.
• Consumer group scales analytics writers horizontally.

Practice next

1. Choose Kafka for event streams needing replay.
2. Design topic per domain: audit, clicks, payments.
3. Set retention and compaction for key topics.
4. Add Avro schema for audit event version 2.
5. Build ksqlDB stream for real-time fraud aggregates.

Remember

Kafka = durable ordered event log. Replay and multiple consumer groups. MeanVerse analytics and audit pipelines.

Compliance audit trail

Regulator requests 90-day transfer activity replay.

Outcome: Reconsume topic from offset into report DB without touching OLTP Mongo.