Maxmemory and Eviction Policies

What you'll walk away with

Explain the core ideas behind Maxmemory and Eviction Policies
Run the sample Redis command or code and verify its output
Apply the technique correctly to the Tutorial Platform and production scenarios

Build the mental model

At maxmemory, Redis evicts keys or rejects writes according to policy. allkeys-lru or allkeys-lfu suits caches; volatile policies select only expiring keys; noeviction rejects writes. Eviction is sample-based, and operating-system headroom is needed for replication and AOF buffers.

Connect it to a real scenario

Test allkeys-lfu on a dedicated catalog-cache instance and measure hit ratio, evicted_keys, and latency under load. Keep sessions on separate capacity and policy to avoid accidental eviction. Alert on eviction and miss rates as well as used_memory versus maxmemory.

Choose policy from data semantics
Separate caches from sessions when needed
Keep headroom and watch eviction metrics

Try the working example

shell

CONFIG GET maxmemory
CONFIG GET maxmemory-policy
CONFIG SET maxmemory 2gb
CONFIG SET maxmemory-policy allkeys-lfu
INFO stats
INFO memory

You should see

You can verify the configured ceiling, policy, and eviction counters.

5-minute try-it

Recommend and justify policies for a pure cache, session store, and mixed workload.

One important caution

A runtime CONFIG SET may disappear after restart unless configuration management is updated too.

Redis — Key Eviction — Redis

Easy traps

A runtime CONFIG SET may disappear after restart unless configuration management is updated too.
Validate sample commands on a local or test instance with recoverable data before applying them to production Redis.

Exercise

Recommend and justify policies for a pure cache, session store, and mixed workload.

You'll know it worked when: You can verify the configured ceiling, policy, and eviction counters.