Live MDX Article Composer

Craft full-featured MDX articles with compliant YAML frontmatter.

Article Frontmatter Metadata

Article Title

URL Slug

Brief / Excerpt

Series Name

Read Time (min)

Tags (comma-separated)

Cover Image URL

How to publish:

Click Download .mdx File and drop it into src/content/ in the blog repository, or copy the code and save directly!

Markdown & MDX Body

160 words

When orchestrating microservices in distributed applications, synchronous REST calls introduce tight coupling and cascading latency.

In this article, we explore how event choreography using Apache Kafka eliminates point-to-point dependencies while guaranteeing at-least-once message delivery.

---

## The Outbox Pattern

To prevent dual-write anomalies between relational database updates and Kafka publishing, use the Transactional Outbox Pattern:

```sql
-- In the same ACID transaction:
INSERT INTO orders (id, user_id, amount) VALUES (1, 'user-101', 250);
INSERT INTO outbox_events (aggregate_id, event_type, payload)
VALUES (1, 'OrderCreated', '{"amount": 250}');
```

A Debezium CDC connector reads the outbox table via Postgres Write-Ahead Log (WAL) and publishes directly to Kafka.

---

## Consumer Group Partitioning

Kafka assigns each topic partition to exactly one consumer within a group, ensuring strict ordering per partition key:

```typescript
const consumer = kafka.consumer({ groupId: 'order-processing-group' });
await consumer.subscribe({ topic: 'orders', fromBeginning: false });

await consumer.run({
  eachMessage: async ({ topic, partition, message }) => {
    console.log(`Received event on partition ${partition}: ${message.value?.toString()}`);
  },
});
```