This is the amount of time data that can be lost and still go back to a recovery point, in other words, this is how often recovery points are taken; in the case of backups, how often a backup is taken (it could be hourly, daily, or weekly), and in the case of a storage cluster, how often changes are replicated.
One thing to take into consideration is the speed at which changes can be replicated, as we want changes to be replicated almost immediately; however, due to bandwidth constraints, the real time replication is not possible most of the time.
Finally, an essential factor to consider is how data is replicated. Generally, there are two types of replication: synchronous and asynchronous.