Cluster replication

What kind of hardware failure do you want to protect against? With a multi-node cluster with shard replicas you're pretty resilient to begin with.

Also, see this recent topic: