What does a replica do in a cluster?

In Redis Cluster, a replica performs several key functions, but on its own it doesn't hold any slots - it copies its master's data.

#1. Copying the master's data

• A replica syncs with a master node and receives a full copy of every slot the master serves.
• Replication is asynchronous: the master doesn't wait for the replica's acknowledgment when writing.

#2. Providing fault tolerance

• If the master fails, its replica is automatically promoted to master.
• The rest of the cluster's replicas and nodes learn about it through Gossip.
• That lets the cluster keep running with no manual intervention.

#3. Scaling reads

• Replicas can be used to serve read requests, taking load off the master.
• Reading from replicas helps scale throughput without increasing load on the primary node.

#4. Important points

• Replicas don't acquire new slots on their own, they only copy their master's data.
• Every replica is tied to a single master and can't serve other masters' slots.
• A replica can have several downstream read clients, but writes always go through the master.

#Summary

A replica in Redis Cluster:

1. Copies its master's data.
2. Provides fault tolerance (auto-failover).
3. Scales reads, reducing load on the master.

It doesn't hold slots itself, but it provides resilience and boosts the cluster's throughput.