# Backups Backups are snapshots of the data and configuration, taken on a schedule you choose or before an operation that could lose data. For a database that means a physical base backup plus the write-ahead log, which is the only way a Postgres restore is actually correct. They are stored inside the same region as the target, which is the one thing that makes restore fast and cheap. ## Take one ``` kraftify backup create --label before-migration ``` ## List and restore ``` kraftify backup list kraftify backup restore ``` Restoring stops the target, swaps the files and starts it again. The target is unavailable for the length of the swap, which is seconds for a small database and minutes for a large one. ## Schedule them A schedule keeps the most recent 24 hourly and 7 daily snapshots and prunes the rest. Pruning is a delete, so the schedule is the only place retention is ever automatic. ``` kraftify backup schedule --hourly --daily ``` > Snapshots are not replicated off the node. A node loss takes its snapshots with it. That is a deliberate cost decision, not an oversight, and it is the main thing we would change first with more funding.