There are several ways to create a backup infrastructure since the amount of data and number of clients varies. Here is how I set mine up at home.
Since the clients we use at home are all Windows, we only need to save a certain set of folders/files. Within each client, we can have multiple user accounts. But since the goal of my backups is to be flexible in what targets the backup can be restored to, we use a combination of methods to backup content.
Within each user account, we specify essential folders that need to be backed up. For me it would be the following folders:
There are two ways I use to back up these folders. I use a sync tool (Synkron) to mirror the current state of the file and folder structure. Backing up the structure allows for restoration where I know all files have been restored to last known good state.
Secondly, I use a file versioning method (Windows 10 includes File History) which basically takes snapshots of the filesystem at set intervals. Any changes that are applied to certain files are updated. The versioning aspect allows for pointing to a specific time to restore to.
The difference between the two are subtle, but help in the the effectiveness of efficient space usage and recovery times.
The main difference between the two is that in synchronization, all deletions are propagated and only 1 version of the file system is retained at any given time. Think of it as a mirrored system where the location of the backup is physically separated (at the very least, different drives). Meanwhile, File History retains all files ever created (if backed up during an interval – more on this later). Deletions are not propagated, and multiple files of similar content are created since the whole purpose is to have versioning of the files.
The reason to use the multiple file backup strategy is for the variation of data access and structure. If the entire drive fails, restoring from the synced backup is a faster solution, while if a file is deleted and the sync job runs, the file is still retained by the File History.
The storage used by both systems varies. Different levels of backups create different levels of time that is required to restore the system. The level of data backup should correlate to the amount of data restore points available for the backup. More on this later.
The last backup method that is part of my backup strategy is to have an image level backup of the system partition. This includes the OS and any programs installed on the drive. The main purpose of this is to restore the OS with programs while retaining the license keys for Windows and Office programs that are subscription based (products activated with Office365 for example).
The amount of retained backups should correlate with the amount of data that is included in the backup. Since Image level backups contain OS and programs, these should be retained the longest but with the restore points that are further apart.
Meanwhile, raw data backups (sync and File History) should be retained for a shorter time.
For example, Image level backups should perhaps have 1 (full) restore point a week (a full backup being retained for ~3-4 backup cycles, with incremental retained for 2 weeks). Therefore, you would have 4 full backups over the span of 4 weeks with 2 weeks of incremental backups.
The combination of retention allows for a reliable restoration attempts in a variety of cases – the changing amounts of files and time of deletion/corruption shall dictate which restoration point to use.