Memcached vs Redis

Memcached and Redis are both in-memory key-value stores used for caching, but they differ significantly in capabilities. Memcached is a simple, multi-threaded cache. Redis is a feature-rich in-memory data structure store.

Design Philosophy

Memcached: Pure Simplicity

Memcached does one thing well: cache key-value strings in memory with LRU eviction. It is multi-threaded, uses a slab allocator for predictable memory management, and has a tiny operational surface. There is no persistence, no replication, no data structures — just fast caching.

Redis: Swiss Army Knife

Redis supports strings, lists, sets, sorted sets, hashes, HyperLogLog, streams, and more. It offers persistence (RDB + AOF), replication, pub/sub, Lua scripting, transactions, and clustering. Redis can serve as a cache, message broker, session store, leaderboard, rate limiter, and more.

Performance Comparison

For simple get/set operations on string values, Memcached and Redis perform comparably. Memcached's multi-threaded architecture gives it an advantage on multi-core machines for pure caching workloads. Redis's single-threaded command processing can be a bottleneck at extreme throughput, though io-threads help with network I/O.

Learn about caching strategies in system design concepts and interview questions.

Memory Management

Memcached's slab allocator pre-allocates memory in fixed-size chunks, which is predictable but can lead to internal fragmentation for variable-sized values. Redis uses jemalloc and per-data-structure memory management, which is flexible but adds overhead for small keys.

The Modern Landscape

Redis has largely replaced Memcached in modern architectures because its additional features (data structures, persistence, replication) outweigh Memcached's simplicity advantage for most use cases. However, Memcached remains relevant for pure caching workloads where simplicity matters.

The Bottom Line

Choose Memcached when you need a simple, multi-threaded cache for string values and nothing more. Choose Redis when you need data structures, persistence, replication, or any feature beyond basic caching. See pricing and system design guides.