====== Redis vs RabbitMQ vs Kafka (Data Structure Comparison) ======

===== 1. Redis =====

Yes — Redis main data structure is:

  * Hashtable (key → value) stored in RAM

==== Key idea ====

  * Very fast lookup: O(1)
  * Works like an in-memory dictionary

==== Mental model ====

  * Redis = key-value store (like a dictionary / hashmap)


===== 2. RabbitMQ =====

RabbitMQ is NOT a hashtable system.

It is a **message queue system**.

==== Core structure ====

  * FIFO Queue (First In First Out)
  * Internally uses queue buffers (Erlang ETS tables + disk storage)

==== How it works ====

  Producer → Exchange → Queue → Consumer

Inside the queue:

  * Messages are stored in order (FIFO)
  * Stored in memory first, then disk if durable

==== Key idea ====

  * RabbitMQ = queue-based message broker
  * Not key-value lookup
  * You do NOT query messages by key


===== 3. Kafka =====

Kafka is NOT a hashtable and NOT a queue.

It is an **append-only distributed log system**.

==== Core structure ====

  * Append-only log file
  * Each topic contains partitions
  * Each partition is a sequence of messages

Example:

  offset 0 → msg1
  offset 1 → msg2
  offset 2 → msg3

==== How it works ====

  Producer → Topic → Partition Log → Consumer reads by offset

==== Key idea ====

  * Kafka = distributed commit log
  * Consumers read using offsets (like file pointers)


===== Comparison Table =====

^ System     ^ Core Structure        ^ Storage Type     ^ Access Style        ^
| Redis      | Hashtable             | RAM              | Key-based lookup    |
| RabbitMQ   | FIFO Queue            | RAM + Disk       | Message consumption |
| Kafka      | Append-only log       | Disk             | Offset-based read   |


===== Simple Mental Model =====

  * Redis    → Dictionary (key-value store)
  * RabbitMQ → Waiting line (queue)
  * Kafka    → Log book (append-only journal)