Redis Cluster

Redis Cluster is a way to run Redis on multiple machines by automatically splitting (sharding) data across them. It provides horizontal scaling — you can add more nodes to handle more data.

Analogy: A single Redis is like one cashier at a store — they handle all customers. A Cluster is like having 10 cashiers — each handles a separate line, so the store can serve 10x more customers.

flowchart LR
    Key["Key: user:42"] --> Hash[Hash the key<br/>CRC16 hash function]
    Hash --> Slot[Compute slot<br/>CRC16(key) % 16384<br/>→ Slot 7623]
    Slot --> Node{Pick node by slot}

    Node -->|Slots 0-5500| Node1[🟢 Node 1<br/>Primary]
    Node -->|Slots 5501-11000| Node2[🟢 Node 2<br/>Primary]
    Node -->|Slots 11001-16383| Node3[🟢 Node 3<br/>Primary]

    Node1 --> R1[🟡 Replica<br/>Backup for Node 1]
    Node2 --> R2[🟡 Replica<br/>Backup for Node 2]
    Node3 --> R3[🟡 Replica<br/>Backup for Node 3]

    style Key fill:#7c3aed,color:#fff
    style Hash fill:#3b82f6,color:#fff
    style Slot fill:#f59e0b,color:#fff
    style Node1 fill:#059669,color:#fff
    style Node2 fill:#059669,color:#fff
    style Node3 fill:#059669,color:#fff
    style R1 fill:#f59e0b,color:#fff
    style R2 fill:#f59e0b,color:#fff
    style R3 fill:#f59e0b,color:#fff

• 16,384 hash slots — Redis hashes every key and assigns it to one slot
• Each node in the cluster owns a range of slots (e.g., Node 1: slots 0-5500)
• When you add/remove nodes, slots are rebalanced (some moved to new nodes)
• Every node knows about every other node — full mesh topology
• Replicas provide high availability — if a primary fails, its replica takes over

# Create a cluster (6 nodes: 3 primaries + 3 replicas)
redis-cli --cluster create \
  192.168.1.1:6379 192.168.1.2:6379 192.168.1.3:6379 \
  192.168.1.4:6379 192.168.1.5:6379 192.168.1.6:6379 \
  --cluster-replicas 1

# Check cluster info
CLUSTER INFO

# List nodes in the cluster
CLUSTER NODES

# Check which slot a key belongs to
CLUSTER KEYSLOT user:42
# Output: (integer) 7623

# Count keys in a specific slot
CLUSTER COUNTKEYSINSLOT 7623

By default, Redis hashes the entire key to determine its slot. Hash tags let you force certain keys to the same slot:

# Keys with {USER} in the same part go to the same slot
SET {USER}:42:name "Alice"
SET {USER}:42:email "alice@example.com"
# Both keys go to the same slot → supports multi-key operations!

• Cluster splits data across multiple Redis servers (horizontal scaling)
• Each key is assigned to one of 16,384 hash slots using CRC16 hashing
• Each node owns a range of slots — add nodes to redistribute slots
• Replicas provide automatic failover if a primary node fails
• Use Cluster when data exceeds one server’s RAM or write throughput is too high
• Hash tags ({...}) let you keep related keys on the same node
• Cluster adds complexity — use only when you truly need it