Redis gets its speed from memory, so an 8 GB server should not automatically receive an 8 GB maxmemory setting. Leave headroom for persistence, replication and operating-system buffers.
Redis documentation recommends leaving memory headroom for buffers when replication or persistence is used. When memory limits are reached, keys may be evicted according to policy or writes may fail under noeviction.
Cache data may be disposable, while losing sessions or queue data can directly affect users. Plan eviction and persistence by role.
Replication buffers, client output buffers and persistence/fork behavior can increase peak memory beyond the dataset.
Start with INFO and config outputs without running dangerous commands in production.
redis-cli PINGredis-cli INFO memoryredis-cli INFO statsredis-cli CONFIG GET maxmemoryredis-cli CONFIG GET maxmemory-policyredis-cli INFO persistenceRedis should communicate with applications over localhost or private networking. Add ACL/auth and TLS as required by the architecture.
A separate VPS becomes useful when Redis memory competes with app/database resources or needs independent scaling/failure isolation.
Low cost
Memory isolation
HA operations
Account for dataset, overhead, buffers and persistence headroom. A fixed recommendation without measuring dataset size is unreliable.
It is the policy that rejects new writes with an error when the memory limit is reached.
Not necessarily for cache-only use; sessions, queues or durable data may require RDB/AOF planning.
Share key count, memory usage, role and persistence needs; we can size the Redis VPS.