In Redis, multiple databases appears to be an idea added in the early days of the project, but was never developed into a fully functional feature, and is not supported in the sharding version of the Redis server, named Redis Cluster. My advice is to not use them. Let all your keys exist in Database 0, the default. If you want to have a Redis server machine (or group of machines in master/slave configuration) that is shared by multiple applications, use key names to keep the data separate. And the same for a server/group that's shared among multiple environments (dev, qa, staging).