The best Backup strategies

Various types of data backup are used as part of a backup strategy. Read on to explore an overview of the most common types. More detailed information can be found in the articles linked below. A backup strategy usually includes periodic full backups as well as incremental or differential backups in between.

Here’s a brief overview of the three main types of backup and their strengths and weaknesses, which we will consider in more depth below.

A low complexity of creation and recovery mean that the backups of this method can be created and restored with on-board means. If a method has a high complexity, special software is usually required. In principle, low complexity is preferable.

The volume of data generated when creating a backup influences two dimensions:

1. Storage requirements of the backup medium: how much space is needed to create another backup?
2. Scope of transfers: backups are copies of data. The copied data must be read from the source, transferred, and written to the destination media. Large transfers require high bandwidth to complete in an acceptable time frame.

Creating a backup takes a little time. This includes the time required for copying the data as well as any other upstream or downstream steps. The time required depends on the complexity of the backup method, the resulting volume of data and the bandwidth available between the source and target system.

Creating backups is only one side of the coin. The effort involved is useless if the backups cannot be restored in the event of damage. In principle, independent backups are preferable to linked backup chains, as this minimises the risk of data loss.

The term `backup´ is ubiquitous in the information age. But what exactly is a backup? The term existed before the widespread use of digital systems. In principle, a backup system is a provision for maintaining operations in the event of an emergency. For example, a hospital has a backup generator. This stands in if the primary power supply fails.

In terms of digital systems, backups are redundantly stored data. Simply put, copies of important data are created and distributed across different media. If the main data storage device is damaged, the data can be reconstructed from one of the backups. Following the 3-2-1 backup rule, one of your backups should be in the cloud. Backups are particularly important in two circumstances:

1. Loss of data. Lost data can be reconstructed from the backup.

Digital systems are virtual. The smallest errors can lead to full data loss. Conceptually, this is like a skyscraper collapsing when a single screw fails. Then the only thing that helps is to have a redundant copy available.

       2. Changing data. Data can be reset to a prior state.

Digital systems develop a “life of their own” as complexity increases. If unintentional changes occur, they are often difficult or impossible to reverse. This is because troubleshooting and correcting errors can consume a great deal of time and effort without guaranteeing a positive result. It is therefore better to restore the data from a previously created copy.

A backup strategy is needed to ensure that backups can be used in an emergency. After all, backups are only valuable if they are planned and created with foresight. A backup strategy includes the following basic considerations:

1. Which data must be backed up?
2. At what intervals?
3. On which media?
4. Using which methods?
5. How is the integrity of the backups ensured?
6. How are backups restored if necessary?

The final points tend to be often neglected. If it takes an emergency for you to notice that your backups don’t contain the desired data or are not suitable for restoring data, it’s already too late. Therefore, it’s important to extensively test the entire process from creating backups to restoring them. A solid backup strategy ensures that there are no nasty surprises in the event of data loss.

There are several methods to create continuous backups of a dataset. The three main backup methods  full backup, differential backup, and incremental backup each offer specific advantages and disadvantages. In this article, we provide an overview of them, but you can read more about each type in our in-depth guides on each backup strategy.

Generally, a backup strategy usually includes several types of backups. First, a full backup is created. This is followed by incremental or differential backups and, if necessary, periodic full backups. Different data sets are backed up with varying frequency using the appropriate methods, depending on requirements.

Let’s assume an organisation has a current dataset of 100 GB in need of backing up. Let’s further assume that the dataset grows by 1 GB a day. As part of a conventional backup strategy, a full backup is created on the weekend. Furthermore, daily amends are to be backed up by another backup. So all backup methods start with a full backup on Sunday. Subsequently, depending on the backup method, only changes are backed up if necessary.

Let’s compare the three types of data backup. First, we compare the size of daily growth in data with data volumes accruing per backup method. The volume of the full backups corresponds to the volume of the data stock. The volume of a differential backup grows linearly over time as of the last full backup. In contrast, the volume of incremental backups corresponds to the volume of data changed:

Creating backups is an important aspect of a backup strategy. But what about data recovery? The table below compares the types of backups by considering the number of backups required to restore them. To restore full and differential backups, a constant number of backups are necessary for recovery. With incremental backups, the number of required backups grows linearly over time since the last full backup:

A full backup creates a complete copy of all data on a system. This can be a copy of all data on a laptop or, for example, a copy of all digital payrolls for the year. If several full backups of the same data are created at different times, there is usually a high level of redundancy in the data they contain. This is because most data only changes at certain points in time. A full backup serves as the basis for the two other types of backups.

Since a full backup covers the entire defined data stock, there is usually a high volume of data to be backed up. It follows that creating a full backup requires a lot of storage space on the backup medium. In addition, the process usually takes a long time. On a positive note, a full backup does not require much effort: you simply back up everything. Furthermore, restoring data from a full backup is usually straightforward.

A differential backup contains all of the data that has changed since the last full backup. The amount of data to be backed up depends on how much time passed between a differential backup and the last full backup. As a rule, however, a differential backup is much smaller and therefore faster to create than another full backup.

To restore data, the differential backup and the last full backup are required. If multiple differential backups are created, they are independent of each other. If one differential backup becomes corrupted, other differential backups are not affected. Some common admin tools allow the creation of differential backups. For example, differential backups can be created with Rsync.

An incremental backup  backs up changes since the last full or incremental backup. The individual backups are small and fast to create. However, incremental backups are not independent of each other. Restoring them requires all incremental backups since the last full backup. If an incremental backup in the chain is damaged, all subsequent incremental backups are worthless.

Due to the small volume of data to be backed up, incremental backups are particularly efficient and therefore widely used. They tend to be used to create frequent backups of smaller changes. For example, incremental backups are useful to back up on a Mac via “Time Machine”. Incremental backups on Windows 10 are easy with the Robocopy Backup tool.