Redis Replication

18. Redis Replication

What is Replication?

Replication creates exact copies of a Redis primary (master) on one or more replica servers. The primary handles all writes and sends updates to replicas, which can serve read requests.

Analogy: Replication is like a teacher (primary) writing notes on a whiteboard while students (replicas) copy everything down. If the teacher is absent, one student can take over.

Replication Architecture

flowchart TB
    ClientWrites[App writes data] --> Primary[🔵 Primary<br/>Read & Write<br/>All writes go here]

    Primary -->|ASYNC replication| Replica1[🟢 Replica 1<br/>Read-only<br/>Backup copy]
    Primary -->|ASYNC replication| Replica2[🟢 Replica 2<br/>Read-only<br/>Backup copy]
    Primary -->|ASYNC replication| Replica3[🟢 Replica 3<br/>Read-only<br/>Backup copy]

    ClientReads[App reads data] --> Replica1
    ClientReads --> Replica2
    ClientReads --> Primary

    style ClientWrites fill:#3b82f6,color:#fff
    style ClientReads fill:#3b82f6,color:#fff
    style Primary fill:#7c3aed,color:#fff
    style Replica1 fill:#059669,color:#fff
    style Replica2 fill:#059669,color:#fff
    style Replica3 fill:#059669,color:#fff

How Replication Works

Replica connects to primary and sends REPLICAOF
Primary forks a child process to run BGSAVE (RDB snapshot)
Primary sends the RDB file to the replica
Replica loads the RDB file (catches up to sync point)
After initial sync, primary streams every write command to replica (replication buffer)
If replication link breaks, replica reconnects and does a partial resync (not full sync)

# Set up a replica (run on replica server)
REPLICAOF 192.168.1.100 6379

# Check replication status
INFO replication

# Make a replica primary again (stop replicating)
REPLICAOF NO ONE

Configure in redis.conf

# On replica server
replicaof 192.168.1.100 6379

# If primary has password set
masterauth your_redis_password

# Replica is read-only by default (recommended)
replica-read-only yes

Benefits

Benefit	Explanation
Read scaling	Distribute read queries across replicas
Data redundancy	Multiple copies of your data
Backup safety	Run backups on a replica (no performance impact on primary)
High availability base	Foundation for Sentinel (automatic failover)

Replication vs Other Solutions

Feature	Replication	Sentinel	Cluster
Read scaling	✅ Yes	✅ Yes	✅ Yes
Write scaling	❌ No	❌ No	✅ Yes (sharding)
Auto failover	❌ Manual only	✅ Automatic	✅ Automatic
Complexity	Low	Medium	High
Typical nodes	1 primary + 1-3 replicas	1 primary + 2 replicas + 3 sentinels	3+ primaries with replicas

In Simple Words

Replication copies data from one primary to one or more replicas
Replicas are read-only — all writes go to the primary
The primary pushes updates asynchronously to replicas
Use replicas for read scaling and backup safety
Replication is the foundation for Sentinel (auto-failover) and Cluster (sharding)
If a replica disconnects, it only syncs the missed commands — not a full resync