Redis vs Memcached

Redis and Memcached are both in-memory data stores used for caching and session management, but Redis has evolved into a multi-purpose data structure server while Memcached remains focused on simple key-value caching.

Overview

Redis is an in-memory data structure store supporting strings, hashes, lists, sets, sorted sets, streams, and geospatial indexes. It offers persistence options (RDB snapshots and AOF logs), pub/sub messaging, Lua scripting, transactions, and clustering. Redis is widely used as a cache, message broker, session store, and real-time leaderboard engine. Memcached is a high-performance, distributed memory caching system designed for simplicity. It stores key-value pairs in memory with an LRU eviction policy. Memcached is multi-threaded, making it efficient at utilizing multiple CPU cores for simple caching workloads.

Key Differences

Data structures: Redis supports complex data types — sorted sets for leaderboards, streams for event logs, hashes for objects. Memcached only supports string key-value pairs with a 1 MB value limit. Persistence: Redis can persist data to disk via snapshots or append-only files, allowing recovery after restarts. Memcached is purely in-memory — all data is lost on restart. Threading model: Memcached is multi-threaded and can utilize all CPU cores efficiently. Redis historically used a single-threaded event loop (with I/O threading added in Redis 6+), which simplifies consistency but can bottleneck on CPU-bound operations. Memory efficiency: Memcached uses a slab allocator that is highly efficient for uniform-sized values. Redis has more memory overhead per key due to its richer data structures.

Pricing

Both are open-source and free to self-host. Managed Redis services (AWS ElastiCache, Redis Cloud) start around $15/month. Managed Memcached (AWS ElastiCache) has similar pricing.

Verdict

Choose Redis for most caching needs — its rich data structures, persistence, and pub/sub make it versatile beyond simple caching. Choose Memcached if you need a dead-simple cache with multi-threaded performance and your workload is purely key-value lookups with no persistence requirements.