Take a look at your footprint, meaning what servers you have employed, and devise a point in time strategy for recovery for each. So, for SQL a point in time recovery system is baked into the SQL Server product, in the form of backups, transaction log backups, etc... For MongoDB, something similar. For server backups people tend to use some type of bit level recovery, which is essentially a point in time recovery type of tool for operating systems. Initially you take a full backup of the server, then you take differentials and incremental backups as well.

Hope this helps.
Bob