The Copy-On-Write optimization process is currently found in the ZFS, Btrfs (Linux) and APFS (macOS) file systems and is an internal file system practice to avoid unnecessary copying and creation of further copies. Specifically, this means that the actual copy will not be fully made until a user changes it. Saving the original file is completely sufficient for the period in which the users involved have not yet made any changes. The copy is initially only created virtually and is delayed when the first adjustment is made and saved. In the file system, this means that changed blocks are not overwritten, but instead are moved (copied) to free space, followed by an update of the references in the metadata. The resulting delay-free creation of snapshots is a further advantage that Copy-On-Write enables transaction-based file systems.