Scaling for Rediss
Scaling Redis means spreading data and load across several nodes. Redis Cluster does this by splitting the keyspace into 16384 hash slots assigned to primary nodes, each with replicas for failover. Clients route commands to the right node, and multi-key operations need keys that share a slot.
Stage: Reliability · Part of the Redis roadmap · Last reviewed 19 November 2026
What to learn in Scaling
- Explain how a key maps to one of 16384 hash slots
- Use hash tags in braces to keep related keys on one node
- Create a small cluster with redis-cli and check cluster state
- Handle MOVED and ASK redirections in a client
- Decide when a single node with replicas is enough instead of a cluster
Scaling interview questions
Why do multi-key commands sometimes fail in Redis Cluster?
Each key belongs to a hash slot hosted on one node, and a command touching keys in different slots cannot run on a single node, so it returns a CROSSSLOT error. Hash tags, such as {user:1}:cart, force related keys into the same slot.
What does a MOVED response mean?
It tells the client that the slot for the requested key lives on another node, giving that node's address. Cluster-aware clients update their slot map and retry there. It is normal after resharding or a failover, and a good client handles it without surfacing an error.
Where Scaling fits in the Redis roadmap
Scaling belongs to the Reliability stage (intermediate level), which takes about 9 hours. Persistence and scale.
Other units in this stage
Back to the full Redis roadmap