How to Build a Stable, High‑Performance Redis Cluster: Persistence, Replication, Sentinel and Sharding

Redis has become increasingly popular, yet developers often encounter concepts such as data persistence, master‑slave replication, Sentinel and sharding clusters without fully understanding their relationships. This article walks through building a stable, high‑performance Redis cluster step by step.

1. Single‑node Redis (Scenario Introduction) – A simple cache architecture where an application reads/writes data to a single Redis instance. When the instance crashes, all traffic falls back to MySQL, causing a massive load spike.

2. Data Persistence (RDB & AOF) – To avoid data loss, Redis can persist data to disk. Two main mechanisms are:

RDB : snapshot of the dataset written periodically by a child process (binary, compressed, fast recovery, small file).

AOF : logs every write command; can be rewritten to shrink size. Provides higher data safety but larger files and slower recovery.

Hybrid persistence : combines RDB snapshot with AOF rewrite (available from Redis 4.0) to get small file size and fast recovery.

3. Master‑Slave Replication (Multi‑Replica) – Deploy multiple Redis instances where one acts as the master and others as slaves that continuously synchronize data. Benefits include reduced downtime (automatic failover) and improved read performance by distributing read requests to slaves.

4. Sentinel (Automatic Failover) – Sentinel instances monitor the master’s health. When a master is deemed faulty, a majority of Sentinels elect a leader (using a Raft‑like consensus) and promote a slave to master, achieving automatic failover without manual intervention.

5. Sharding Cluster (Horizontal Scaling) – When write traffic exceeds a single master’s capacity, multiple Redis nodes are organized into a cluster. Data is partitioned across nodes using a routing rule (key‑to‑node mapping). Two sharding models are discussed:

Client‑side sharding : the application (or a library) determines the target node.

Server‑side sharding (Proxy) : a proxy layer handles routing, keeping the client unaware of the cluster topology.

6. Summary – The article recaps the evolution from a single‑node cache to a robust Redis cluster: persistence (RDB/AOF/hybrid), replication for high availability, Sentinel for automatic failover, and sharding for horizontal scalability. By combining these techniques, developers can achieve long‑term stability and high performance for Redis‑backed services.