Understanding Redis 3 Cluster: Key Concepts, Features, and Operations

Redis 3 officially supports clustering, addressing many inconveniences of building a Redis cluster.

Previous Cluster Issues

Redis can be imagined as a box holding N {key,value} pairs. When the box fills up, additional boxes are added, potentially many. Determining which box a new key belongs to and locating a key later requires a rule, such as using the hash value modulo the number of boxes.

This calculation was previously manual and cumbersome, especially when adding or removing Redis nodes.

Redis 3 Cluster Features

Operate keys as if on a single Redis instance without worrying about their node location.

Add or remove Redis nodes online without stopping the service.

Allocate different numbers of keys to nodes dynamically based on node performance.

Each node has a slave with automatic monitoring and failover.

Core Idea of Redis 3 Cluster

The cluster introduces the concept of hash slots . There are 16384 slots. Keys are not placed directly into Redis; they are first assigned to a slot, and slots are mapped to Redis nodes.

This can be visualized as placing keys into small boxes (slots) first, then placing those boxes into larger boxes (Redis nodes).

To locate a key, compute its hash to find the corresponding slot, then use the slot‑to‑node mapping to identify the node that holds the slot, and finally perform the operation on that node.

Operation Examples

Read/Write a Key

Add a Redis Node

Assume three existing nodes each handle roughly a third of the slots (16384/3 ≈ 5461). To add a new node, redistribute some slots from each existing node to the new one. The number of slots per node can vary based on node performance.