Memcached

High-performance, distributed in-memory key-value store for caching database queries, API responses, and rendered content. Alleviates database load by keeping frequently accessed data in RAM and serving it with sub-millisecond latency. Free and open source, built for simplicity and speed rather than persistence or advanced data structures.

Widely adopted at scale by platforms like Facebook, Wikipedia, and Reddit. Differentiates from Redis by focusing solely on caching: no persistence, no pub/sub, no complex types, which keeps the codebase lean and operations predictable. Client-side hashing distributes keys across servers; each node operates independently with no cross-talk or replication. APIs and client libraries exist for most popular languages.

Key capabilities:

• O(1) operations with lock-friendly design for predictable low latency
• LRU eviction and per-item TTL for automatic cache management
• Horizontal scaling: add servers to increase aggregate memory, clients hash keys to nodes
• Simple ASCII protocol, including telnet for debugging and introspection
• BSD-licensed, runs on Linux, BSD, and macOS

Use cases: Offloading read-heavy database workloads, caching session data or API responses, reducing repeated computation for dynamic web pages. Developers integrate it as a drop-in cache layer with standard client libraries, use the telnet interface for quick inspection, and rely on TTL plus LRU to avoid manual cache invalidation in many scenarios.