Differential Backup

We can tirelessly speak about the necessity of information backup. But, unfortunately, the strongest motivation for taking backup seriously arises mostly after incidents of serious loss of critical data.

Backup creation methods

The easiest and clearest method of backup copying is so called Full Backup.

Differential Backup

The essence of this method is taking all our data and making its physical copy. As soon as Full Backup is completed we will have two copies of our entire data: one is the original, and the other is the backup. Both copies will be identical. The downside is that full copy will occupy the same amount of disk space as the original; besides backup creation process can take a considerable period of time.

Due to this reason, two other methods of backup are used more often: Differential backup and Incremental backup. These two types introduce a combination of periodic, but not frequent, backup creation, and creation of differential or incremental copies in between of full copies once an hour, for example. Differential and incremental backups are created quickly and consume significantly less disk space compared to full copies, since they do not include all files, but only those created or changed after the latest full backup.

Differential backup
Let us examine the method of backup called Differential backup in detail.

When it is used, only those changes are backed up which occurred after the latest Full Backup. Since differential backups can be created either on image level or file level, this set of changes will be presented in a form of altered disk blocks (for image-level backup) or changed files collection (for file-level backup).

Major benefit of differential backup lies in significant decrease of time needed to create the backup compared to Full Backup. On the other hand, recovery after a failure takes more time. Recovery process will require two data restoration operations to be performed. During the first operation, the data from Full backup will be recovered, while the second will recover data from differential backup.

When used for inexpensive data storage subsystems, file-level Differential backup is selected in cases when applications create numerous small files and alter only small subset of those files after Full backup is created. However, this backup method is not used in case the hard disk drive is used by Database management applications, which constantly make changes in huge database files. Thus, in case of file-level backup, a backup copy of the huge will be created over and over. Microsoft Exchange is an example of a program which tends to constantly alter the contents of huge database files.

Examples of Differential backup implementation
Let’s consider a few real life examples to demonstrate how Differential backup technology is implemented in practice:

Files and directories. Suppose we need to back up a directory, which contains 50 files having the size of 1 MB on average. Accumulated changes in files comprise 2 MB per day. Creation of backups of all changed files will require 50 MB of disk space per day. Using differential backup will allow decreasing this amount to 2 MB, then 4 MB, then 6 MB, that is no more than 2 megabytes per day, exactly the amount taken by changes themselves.

Disk images. Differential backup is very efficient to create a disk image and allows saving disk space. However, to compute the difference Full backup still needs to be created, and thus it requires corresponding volume of free disk space, which is not an option for many users.

Databases. Backup of databases is the sphere, where Differential backup is extremely important. A typical backup plan used by modern companies requires a weekly backup of Exchange Server. Instead of copying tens of gigabytes of data, Differential backup allows decreasing backup sizes to hundreds of megabytes and only make backups of new correspondence, which appeared after the latest Full Backup took place.