Redis persistence: RDB snapshots vs AOF logging explained

Persistence in Redis

Redis is in-memory but can persist data to disk so it survives restarts. Two mechanisms exist; you can use one or both.

RDB — point-in-time snapshots

Redis forks a child process and writes a compact binary snapshot to dump.rdb. Triggered automatically by rules in redis.conf or manually with BGSAVE.

# redis.conf — save if 100 writes in 60 seconds
save 60 100

# Manual snapshot
redis-cli BGSAVE

RDB is fast to restore but you lose writes since the last snapshot.

AOF — append-only file

Every write command is appended to appendonly.aof. On restart, Redis replays the file to reconstruct state. More durable, larger file, slightly slower writes.

# redis.conf
appendonly yes
appendfsync everysec   # fsync once per second (recommended)

Choosing a strategy

Use RDB alone for caches where some data loss is acceptable. Use AOF (or both) when you cannot afford to lose writes. Set appendfsync always for maximum durability at the cost of throughput. Set appendfsync no to let the OS decide — fastest but least safe. For ephemeral caches, disable both and add save "" to the config.