NotesSystem
Redis
Also: Valkey
An in-memory data-structure server (strings, hashes, lists, sets, sorted sets, streams) that runs commands on a single main thread. It's the most common shared cache tier, and it's also used as a queue, rate limiter and lock service.
Also in: Caching › In the real world
Design choices
- Caching: in-memory, with optional persistence (RDB snapshots, AOF log) and replication. Caches usually run it with persistence relaxed.
- Cache Eviction Policy:
maxmemory-policypicksallkeys-orvolatile-LRU, LFU, random, LRM (least recently modified, since 8.6), orvolatile-ttl. All are approximated by sampling (5 keys by default, fed into a pool of candidates), and LFU uses a decaying probabilistic counter. The default isnoeviction, which rejects writes when full. - Cache Invalidation: per-key TTLs, plus server-assisted client-side caching ("tracking") that pushes invalidation messages to clients holding local copies.
- Licensing: moved to source-available licences in March 2024. The Linux Foundation's BSD-licensed Valkey fork followed days later, and Redis 8 added AGPLv3 as an option in 2025.
Notable trade-offs
- The single-threaded command model makes each command atomic, but one slow command (a huge
KEYSorDEL) stalls everyone. - Rich data structures invite using it as a database. Persistence and failover are weaker than a real database's.
Connections
- Related
- Actual Serial Execution