Full backups
A full backup exports the complete VM, all of its disks and its configuration, to a backup repository (BR) on every run. Each run produces a single, self-contained XVA archive: no chain, no dependencies between runs. The mental model is simple, and so is the restore, since any one archive is enough to bring the whole VM back.
Full replication (DR)
Full replication is the current name of the job type formerly called Disaster Recovery (DR). Both names describe the same thing, and you will still meet the old one: XO 6 labels this mode Full replication, while the XO 5 interface still labels it Disaster Recovery, and replicas keep a Disaster Recovery tag.
Incremental backups
_NOTE:_ Formerly known as Continuous Delta backups
Incremental replication (CR)
Incremental replication is the current name of the job type formerly called Continuous Replication (CR). Both names describe the same thing, and you will still meet the old one: XO 6 labels this mode Incremental replication, while the XO 5 interface still labels it Continuous Replication, and replicas keep a Continuous Replication tag.
Metadata backups
Every XCP-ng or XenServer pool keeps a database, the XAPI database, describing everything that isn't the data itself its configuration, holding users, authorizations and resource sets, backup jobs, and plugin settings.
Mirror backups
The goal is to replicate a backup from one backup repository (BR, formerly called remote) to another. For instance, you make your backup to in-house NFS storage, and then replicate to bigger, slower and cheaper storage with a longer retention.
Rolling snapshots
Snapshots are NOT backups. A snapshot lives on the same storage as the original disk: if you lose that disk or the whole storage repository (SR), you lose every snapshot with it. Rolling snapshots are a convenient way to roll back in time, not a way to survive a storage failure.