LogScale's disaster recovery abilities rely on the persisted data in bucket storage being replicated to another location. Internal metadata (the global database) and segment files are continuously written to bucket storage and replicating the object storage to the other location allows operators to create a new cluster using the replicated bucket as the source of data for the cluster. This process requires starting one instance with several configuration options set to enable the recovery and rewriting of internal metadata. See Starting a New LogScale Cluster Based on Another with Buckets for more information.