Skip to main content

What does Redis Cluster consist of?

RedisJunior~2 min read
Suggest an edit

Redis Cluster consists of several types of nodes and support mechanisms that together provide scaling, fault tolerance, and automatic recovery.

1. Master nodes

  • The main working nodes, which hold the data.
  • Every master is responsible for a specific range of hash slots (out of the 16384 possible).
  • They handle writes and reads.

Example:

javascript
master1 → slots 0-5460 master2 → slots 5461-10922 master3 → slots 10923-16383

If a master fails, one of its replicas automatically becomes the new master.

2. Replica nodes

  • Copy a specific master's data and act as a backup.
  • Normally they only read data, but on their master's failure, they perform automatic failover and become the new master.
  • Replication is asynchronous.

Example:

javascript
master1 → replica1 master2 → replica2 master3 → replica3

3. The data-distribution mechanism, hash slots

  • Every Redis Cluster key is distributed across 16384 hash slots.

  • Every master is responsible for a portion of those slots.

  • The distribution is computed with this formula:

    javascript
    CRC16(key) % 16384

  • That lets the cluster know exactly where every key is stored.

4. The Gossip protocol (how nodes interact)

  • The cluster's nodes exchange state with each other through a built-in Gossip protocol.
  • Every node regularly tells the others:
    • what it is (master/replica),
    • which slots it serves,
    • which nodes are available or unavailable.

That lets Redis Cluster operate with no central managing server, the system self-coordinates.

5. The failover mechanism

  • If a master is unreachable, its replica gets confirmation from a majority of the other masters and becomes the new master.
  • The other nodes learn about it through Gossip and update their metadata.

6. Cluster-aware clients

  • Redis clients (drivers) need to understand MOVED and ASK messages.
  • When a key lives on a different node, the client is automatically redirected to the right master.

Redis Cluster's overall structure

javascript
master1 → replica1 master2 → replica2 master3 → replica3 ↓ (every node is connected via Gossip) ↓ data is distributed across 16384 slots

Summary: Redis Cluster consists of:

  1. Several master nodes, holding the data.
  2. Their replicas, providing fault tolerance.
  3. The hash-slot mechanism, distributing the keys.
  4. The Gossip protocol, connecting every node.
  5. Failover and client-side routing, keeping the system running continuously and letting it scale.

Short Answer

Interview ready
Premium

A concise answer to help you respond confidently on this topic during an interview.

Suggest an edit