Your Redis workload has outgrown a single node. You need to scale horizontally and still perform multi-key operations like MGET on related keys (for example, user and cart for the same customer). What should you do?

The correct solution is to create a cluster-enabled Azure Cache for Redis with multiple shards and refactor key names using hash tags. Redis Cluster distributes data across multiple nodes (shards), enabling horizontal scaling. Hash tags (e.g., {123}) ensure that keys with the same tag are stored in the same hash slot, allowing multi-key operations like MGET to work correctly across related keys even in a clustered environment. Scaling up to a larger non-clustered cache (option 2) only provides vertical scaling and eventually hits limits, failing to address the need for horizontal scaling. Distributing keys round-robin across multiple client connections (option 3) would scatter related keys across different cache instances, making multi-key operations impossible without complex client-side logic. Enabling Premium geo-replication (option 4) is for disaster recovery and global distribution, not for horizontal scaling of a single workload or enabling multi-key operations across shards.