Partitions

Introduction

What is Partitions? Partitions are ordered append-only logs.

Understanding the topic

What happens — Partitions:

• Each partition is an ordered append-only log.
• Producer key picks the partition.
• Ordering is guaranteed within one partition only.
• More partitions enable more consumer parallelism.

Visual explanation

Pipeline view:

Producer
  ↓
Topic partition log
  ↓
Consumer group A
  ↓
Consumer group B
Replay is possible because records are retained

Step-by-step explanation

1. Key record — Producer sets key or round-robin.
2. Hash — partition = hash(key) % count.
3. Append — Leader writes to log segment.
4. Assign — One consumer per partition in group.

Informative example

Example:

kafka-topics --bootstrap-server localhost:9092 --create --topic partitions --partitions 6 --replication-factor 3
kafka-console-producer --bootstrap-server localhost:9092 --topic partitions
kafka-console-consumer --bootstrap-server localhost:9092 --topic partitions --from-beginning
kafka-consumer-groups --bootstrap-server localhost:9092 --describe --group partitions-service

Execution workflow

Best practices

• Design events around business facts: OrderCreated, PaymentAuthorized, InventoryReserved.
• Pick keys based on ordering and load distribution.
• Document owner, retention, schema, partitions, and consumers for every topic.
• Assume duplicate delivery and build idempotent consumers early.

Common mistakes

• Expecting global ordering across a topic.
• Creating one partition because local demos work.
• Skipping schema governance until multiple teams deploy independently.

Summary

Partitions — Start every design with event ownership, topic naming, key choice, schema contract, retention, replay plan, and monitoring.