Memcached versus Redis

Compare Memcached and Redis by the work your application needs the cache to perform: storing opaque values, modifying data structures, recovering state, routing requests, and handling failures. Product version and deployment mode matter as much as the product name.

Introduction

Both systems support client-server access to in-memory data. Actual latency depends on commands, payloads, network paths, contention, and persistence settings; “sub-millisecond” is not a guarantee for every deployment.

Memcached

Memcached stores values as opaque byte sequences under keys. Applications serialize complex objects and interpret the returned bytes. In a classic client-routed deployment, cache servers operate independently: the client selects the target, while the server performs lookup, storage, expiry, and eviction.

Hash-table lookup has expected $\text{O(1)}$ computational cost. That is not deterministic end-to-end query latency: payload copying, network transfer, memory pressure, and concurrent work still matter. Benchmark the request mix and topology you plan to run.