Replication

Why Replicate?

If a cache node crashes, all data on it is lost. Replication copies data to backup nodes.

Primary Node: user:123 = "John"
    ↓ replicate
Backup Node 1: user:123 = "John"
Backup Node 2: user:123 = "John"

How It Works

1. Client SET key value
2. Primary node stores locally
3. ReplicationManager sends to N-1 backup nodes
4. Backups store the same key-value
5. If primary crashes → backup serves the data

Quorum Write

Write succeeds when majority of nodes acknowledge:

3 nodes: write succeeds if 2/3 nodes confirm
5 nodes: write succeeds if 3/5 nodes confirm

Trade-offs

	No Replication	With Replication
Write speed	Fastest	Slower (network)
Read speed	Fastest	Fast
Fault tolerance	None	High
Memory usage	1x	2-3x

Implementation

import { ReplicationManager } from "distributed-cache";

const repl = new ReplicationManager({
  replicationFactor: 2, // copy to 2 backup nodes
});

// When primary stores a key
repl.replicate("user:123", userData, ["backup-1", "backup-2"]);

// When key is evicted
repl.untrackKey("user:123");