Understanding Redis Cluster: Sharding, High Availability, and Hash Slot Mechanics

1. Redis Cluster Overview

1.1 Automatic data sharding, each master holds a portion of data

1.2 Built‑in high availability; the cluster continues operating when some masters fail

In a Redis Cluster each node opens two ports, e.g., 6379 and 16379 (6379+10000). The second port is used for the cluster bus, which handles node communication, failure detection, configuration updates, and failover authorization.

The cluster bus uses a binary protocol optimized for efficient data exchange, consuming less bandwidth and processing time.

2. The oldest hash algorithm and its drawbacks (massive cache rebuilds)

3. Consistent hashing with virtual nodes for automatic cache migration and load balancing

4. Redis Cluster hash slot algorithm

Redis Cluster defines 16384 hash slots. For each key, CRC16 is computed and modulo 16384 yields the slot.

Each master owns a subset of slots; for example, with three masters each may hold about 5000 slots.

Hash slots simplify node addition and removal: adding a master moves some slots from existing masters; removing a master transfers its slots to others.

The cost of moving hash slots is very low.

Clients can force keys to the same slot using hash tags.