Memcached

Memcached is a high-performance, distributed in-memory key-value cache designed in 2003 by Brad Fitzpatrick at LiveJournal. It is the spiritual ancestor of Redis — simpler, multi-threaded, and laser-focused on one task: caching opaque byte strings keyed by short string keys. Memcached has no persistence, no replication, no data types beyond byte strings, and no cluster mode — deliberately, because its design goal is the smallest possible cache primitive.

Key Features:

• Multi-Threaded. Unlike Redis’ single-threaded model, Memcached scales to many cores per instance — a single node can saturate a 10 GbE NIC.
• LRU Eviction. When memory fills, oldest-accessed items are evicted automatically — pure cache semantics.
• Sliced Memory Allocator. Slab classes prevent the fragmentation that pure malloc would cause for variable-size values.
• Client-Side Sharding. No server-side cluster — clients use consistent hashing to spread keys across nodes.
• Tiny Footprint. The whole server is < 5,000 lines of C; trivially deployable on any platform.

Memcached vs. Redis:

• Choose Memcached when you need a pure ephemeral cache, want multi-core throughput per instance, and don’t use any of Redis’ data structures, persistence, or pub/sub.
• Choose Redis for everything else — rate limiting (sorted sets), session state with persistence, leaderboards, queues, pub/sub, distributed locks, vector search.
• For most teams in 2026, Redis is the default; Memcached is a niche choice for pure-cache fleets at very high QPS where its simplicity is a feature.

Use Cases:

• Object cache for high-traffic LAMP / Rails / Django applications — the original use case.
• Edge / sidecar cache where simplicity matters more than feature richness.
• Multi-tenant cache fleets where per-instance multi-core throughput dominates total cost of ownership.
• Legacy systems that have run Memcached for 15+ years and haven’t had a reason to migrate.