Redis and similar.
WHAT THEY ARE
Fast stores holding data in memory, addressed by key.
WHAT THEY ARE USED FOR
Caching expensive results Sessions Queues Rate limiting Temporary counters
WHY THEY ARE FAST
Memory, and simple operations.
WHAT THEY ARE NOT
A replacement for a database.
WHY
Memory is limited and expensive, and durability guarantees are weaker.
WHAT PERSISTENCE OPTIONS EXIST
Periodic snapshots An append-only log of changes
WHAT TO CHOOSE
Depends on whether losing recent data matters.
WHAT TO ALWAYS SET
A memory limit, and a policy for what happens when reached.
WHY
Without one, it consumes memory until the machine suffers.
WHAT POLICIES EXIST
Evict least recently used keys Evict only keys with an expiry Refuse writes
WHAT TO SET EXPIRIES ON
Everything cached.
WHY
Cache without expiry becomes stale data with no end.
WHAT TO NEVER EXPOSE
The service to the internet.
WHY
Many deployments have no authentication by default, and they are actively scanned.
WHAT TO CONFIGURE
Binding to a local address, a password, and disabling dangerous commands.
