Home > Glossary > Enterprise Backup: Types, Differences and Policies. Full, Incremental, Differential and Synthetic Full Backups.

Enterprise Backup: Types, Differences and Policies. Full, Incremental, Differential and Synthetic Full Backups.

Updated 14th September 2022, Rob Morrison

There are quite a lot of stories of businesses getting in trouble because of the data they lost one way or another. Some couldn’t even continue working after they lost it. At the same time, more and more people are saying that such accidents are preventable through data backups in the cloud services.

For example, you tested some solutions for cloud backup, chose the one you think suits you the most, and started reserving the data you’re working with.That way, even if there’s an accident of some kind, you’ll be able to recover most of the data lost, depending on how much time has passed since the last backup happened.

But are you sure you know the differences between types of backup used? Essentially, there are three “main” types generally recognized by everyone, and there are some newer types and variations of existing ones. Next we’ll go through them, one by one.

Full Backup

The most basic backup type existing nowadays. Everything that you have is completely backed up the first time you enable it. And when the next backup time is due – all of the data is backed up again.

The upside of it is that you always have an easy way to completely recover your data, and you have it all in one dataset. On the other hand – full backups require a lot more storage space and bandwidth than other types because of their size. The whole process of backing up all of your data is quite slow as well.

To compensate that, in many cases full backup is used as the initial one, followed by other, less demanding types of backup, like differential or incremental. There’s also the option to perform full backups with some intervals (be it once or twice a week, once a month, or any other interval). That info, and many more, is something that’s usually described in company’s backup policy.

A backup policy is an easy way to specify who’s responsible for your backups, what kinds of backups are used, how often they’re performed, etc. Other advantages of having a good backup policy include a clarification of all the responsibilities, policies and procedures regarding data backup (with the ability to change them as well); identification of any other existing procedure or policy and if any of them interact with the backup policy in any way; specifies the exact schedule of backups performed; makes sure that more than one copy of your data exists in different places to prevent complete data loss due to hardware or software failure and others.
Backup policy also includes information about specific types of backup used at a time and ensures that the saved data can be restored when needed, specifying the person responsible for that exact process.

The most common practice for full backups is to perform them once a week, with some exceptions, like new application being installed, or server OS or database being upgraded. Big intervals between backups are not recommended due to the amount of data lost in the event of disaster. This is also why there are other types of backup, for example, an incremental one.

Incremental Backup

Incremental backup is the second of the most common backup types, and it works quite differently. The main purpose of this backup is to preserve all of the data you’ve changed since the last time a backup was done, no matter the type of it. Next you’ll see an example to help with understanding the concept.

Let’s say you’ve done a full backup Monday night. When you attempt to run an incremental backup at Wednesday – it’ll backup all of the files changed since Monday. Additionally, when the next incremental backup process is launched (for example, Thursday) – only the files changed after Wednesday’s backup would be saved. The primary goal of incremental backup is to reduce the time between backups, with less data to be backed up each time.

The primary advantage of this type of backup is the shortened backup intervals, and, additionally, less data to be preserved each time. This method also allows for keeping and redeeming multiple versions of the same data. However, it has its own share of disadvantages, like longer recovery time, the strict requirement of full backup already performed for the incremental backup to begin, the added possibility of some data being lost if either full or incremental backup is missing or corrupt. Also there’s the fact that finding a specific file in the backed up data is harder because there is more than one dataset that needs to be searched.

Differential Backup

Incremental backup isn’t the only backup type that works in tandem with full backup. The other one is called differential backup. The main difference between this one and an incremental backup is that differential backup preserves any data changed since the last full backup every time it’s launched.

Let’s work with the same example. The starting point is a full backup made on Monday night. Then you launch a differential backup on Wednesday – it creates a copy of any changes made since Monday. But when this type of backup is launched again on Friday – it backs up everything that was changed since Monday and Wednesday. The long and short of it is that, essentially, differential backup is a number of incremental backups applied cumulatively.

This type of backup still requires less storage space than the full one, and, same as the incremental, it allows for keeping multiple versions of the same file. The added benefit is that you only need the latest differential backup and the full backup for data restoration to work perfectly. It also shares a number of disadvantages of the incremental type. A full backup is still required to begin any number of differential backups, you still need to have both the said full backup and the differential backup to launch the recovery process. And there’s also the fact that differential backup is slightly slower than incremental backup.

It is worth noting that those three aren’t the only backup types that exist. Next you’ll see a number of less known and/or more modern types of backup that a lot of people haven’t even heard of.

Synthetic full backup

One of those is synthetic full backup. Like most of backup types, it requires a full backup to start. Then it creates a number of incremental backups, and at some point all of the incremental backups are integrated to the existing full backup, rewriting older data and synthesizing more up-to-date full backup. After that – the process begins anew.

This is one of the few backups that have much less storage requirements than the others because the data gets overwritten. The disadvantage of that is the unavailability to keep several versions of the same file.

Modification of this type of backup is also possible. For example, you could use a variation of synthetic full backup called progressive virtual full backup. The difference is quite simple – virtual full backup uses advanced index management to severely decrease the data needed for the whole backup process since instead of writing full backup the second time this system uses advanced index management to decrease the size of the newly created “full” backup to just a fraction of its original size. And it saves quite a lot of time, too (the amount of data that needs to be saved is significantly lower).

Reverse Incremental backup

The other variation of incremental backup type is the reverse incremental backup. As the name says, the process is pretty much reversed, the only thing that’s still fixed is the requirement of full backup to begin the process. After that, reverse incremental creates only incremental backups of data at certain points of time, but each of the backups made is immediately fused with the full backup, making a new full backup with updated data. Additionally, the incremental backups taken are kept “behind” the full backup to create the possibility of rolling back to the previous version of your database.

Reverse incremental backup is also negating one of the disadvantages of the original incremental backup – with reverse incremental there’s always a recent full backup at hand, so there’s no need for every incremental backup to make a complete recovery. That’s why reverse incremental is favored over usual incremental type most of the time.

Forever Incremental backup

Another variation of incremental backup is called forever incremental. It starts with a full backup, as usual, but the process differs after that. This system creates a number of data “blocks” each time a backup process is launched, and deletes older “blocks” as newer ones are created.

Same example – Monday is the starting point, and the backups are done daily. A full backup is done on Monday, and two new “blocks” are created (A and B) on Tuesday.

That way on Wednesday block A is deleted and block C is created instead. The same process is done on Thursday – block B is deleted and block D is created, and so on. This system works on deleting all of the duplicated data on a fixed-time basis, keeping storage requirements lower than usual. This type also decreases the bandwidth requirements because of the duplicate data being deleted frequently.

Mirror backup

Not all of the backup types are variations of incremental backup type. For example, there’s mirror backup. As the name suggests – it is an exact copy of your data, stored separately. The difference between this and the full backup type is that mirror backup applies no compression and deduplication to your data, basically allowing you to access this copy whenever you want to do so.

The obvious drawbacks of this method are the fact that due to the data not being compressed you’ll need a lot more storage space than with usual backups, and due to the specific nature of mirror backup type you can’t store any other version of the data except the most recent one.

Continuous backup

One more backup type that is somewhat different from the others is the continuous backup. It’s also called “real-time” backup, or Continuous Data Protection (CDP). Whenever the system detects a change in data – the backup of a changed part is made. Also this type of backup is able to speed up the data restoration process up to mere seconds if the physical disk storage is used while maintaining this backup type.

There are two variations of this backup type: true CDP and near CDP.
True CDP is copying all of the changed data the moment there’s a change in any of the monitored data. Near CDP does the same but with preset time intervals.

Also worth noting that there are two main ways of implementing this backup type:

copying all the data on internal servers or some sort of hardware storage;
copying all that outside of your typical data path (third party storage services and such).

Some services even provide the ability to use both of them at the same time for better data protection.

Other than that – it works the same as the usual incremental type: the first backup is the full one, and the consequent ones are incremental, created every time a change is made. It is one of the more secure methods of preserving your data because it is monitored in real time.

Conclusion

That was an overview of the most of the backup types, with some less common ones described. But the main difference between them should be clear enough: at the core of every backup is a full type of backup. You can use only this one, with expected drawbacks, or you can choose one or several others if they better fit your needs. The most important thing of them all is that no matter which one you choose and how much it suits your needs – any backup is better than having no backup at all and losing your data if or when any sort of problem arises.

About the author

Rob Morrison is the marketing director at Bacula Systems. He started his IT marketing career with Silicon Graphics in Switzerland, performing strongly in various marketing management roles for almost 10 years. In the next 10 years Rob also held various marketing management positions in JBoss, Red Hat and Pentaho ensuring market share growth for these well-known companies. He is a graduate of Plymouth University and holds an Honours Digital Media and Communications degree, and completed an Overseas Studies Program.