November 6, 2017
Backup Types Explained: Full, Incremental, Differential, Synthetic, and Forever-Incremental
VM backup solutions for VMware and Hyper-V environments differ in many aspects, which can be the subject of a lengthy article. In this article, we are going to touch upon only one of these aspects – backup techniques, which are generally known as backup types or methods. Below is a brief overview for each of the traditional (i.e., full backup, incremental backup, and differential backup) and newly-emerged (i.e., forever-incremental and synthetic) backup types, as well as variations thereof (i.e., mirror, reverse incremental, smart, and continuous backup), so you can see how they differ and what advantages each of them offers.
As you may guess from the name, a full backup creates a complete copy of source data set. This backup type is considered to be the best data protection option in terms of simplicity and speed of recovery. However, due to the large volume of data to be copied, the full backup is a highly time-consuming process (it can take up to 10X more time, as compared with other backup types). It also imposes considerable workload on the network each time a backup job is run, thus, interfering with routine operations of your infrastructure. Besides, continuously arriving full backups consume a lot of storage space in the backup repository.
That is why most organizations make full backups only periodically, combining them with other backup types. Thus, a full backup often becomes a starting point in using other backup types.
To summarize, the advantages of making full backups are:
- Fast data recovery in case of disaster;
- Better storage management, since the entire data set is stored in a single backup file.
However, using the full backup type has quite a lot of disadvantages:
- Running full backups takes large amounts of time;
- You need to have a large-capacity storage to keep all of your regularly created full backups;
- Since each full backup file contains the entire set of your data (which is often confidential), your business may be put at stake if this data is illegally accessed by an unauthorized person. However, this risk can be avoided by data encryption, if your backup solution provides for such a feature.
Incremental backups serve to reduce the amount of time and network workload needed to run consecutive full backups. The starting point for making an incremental backup is making one full backup first, and then only those blocks of data that have changed since the last backup job are backed up in increments. Depending on the backup retention policy, after a certain period of time, a new full backup is made to repeat the cycle.
To illustrate the incremental backup process, suppose you made a full backup on Monday leaving the rest of the week for incremental backups. Thus, on Tuesday, only data that changed since the full backup is backed up. On Wednesday, only data that changed since Tuesday is backed up and so on. So, incremental backups can be run almost as often as desired, since each time only the most recent changes, if any, are backed up and stored in the backup repository.
Incremental backups are fast and require much less storage space, as compared to the full backup, but the recovery process is more time-consuming, since you need to restore both the latest full backup and the whole chain of consecutive increments. If one increment in the chain is missing or corrupted, it will be impossible to perform full recovery.
The advantages of incremental backups are:
- Backup jobs run at high speed, as only increments are backed up;
- Less storage space is required;
- Can be run as often as desired, each increment being an individual recovery point.
Disadvantage of incremental backups are:
- Slow full restore of data, as you need to restore both the initial full backup and all subsequently created increments;
- Successful recovery of data depends on the integrity of all increments in the chain.
The differential backup type is an intermediary solution between the full backup and incremental backups. Similar to the incremental backup, the starting point for the differential backup is making a full backup first and backing up only the changed data afterwards. However, in contrast to incremental backups, the differential backup saves not the data that changed since the most recent backup, but the data that changed since the initial full backup. Therefore, the full backup is a constant reference point for sequential backups. Thus, the differential backup allows to restore data faster, as compared to the incremental backup, since it requires only two pieces of backup – an initial full backup and the latest differential backup.
In terms of backup/restore speed, the differential backup type is somewhere between the full and incremental backup types:
- Its backup operation is faster than that of the full backup but slower than that of incremental backups;
- Its restore operation is slower than that of the full backup but faster than that of incremental backups.
The storage space required for differential backups is, at least for a certain period, smaller than that needed for the full backup and bigger than that necessary for the incremental backup. The tricky side is that as time progresses, increments of the changed data sets may grow, and with every backup they may take up even more storage space (and time) than regular full backups.
A mirror backup is similar to the full backup. This backup type creates an exact copy of the source data set, but only the latest data version is stored in the backup repository with no track of different versions of the files. In contrast to other backup types, all individual backup files are stored not in a single compressed/encrypted container file, but separately, just as they are in the source. This allows for getting direct access to backup files without performing a restore operation. The source data is “mirrored” by the mirror backup file. Further on, the mirror backup copies only modified files.
It’s needless to say that this backup type, though advantageous in certain aspects like fast recovery or convenience of direct access to individual files, has its drawbacks: high storage space requirements, high risk of unauthorized access and data corruption or misuse.
Another important point you need to be aware of is that you should use mirror backups with caution. When a file in the source is deleted, the same file in the “mirror” is also deleted. This implies that any adverse modifications occurred in the source due to human error, accident, sabotage, or malware action may cause the same effect in the backup.
Reverse Incremental Backup
The reverse incremental backup is similar to other backup types and starts with running a full backup. After that, incremental backups are made with each successive incremental backup reversibly “injected” into the full backup, thus synthetizing a full backup, which is the latest version of the data set. Moreover, all incremental backups applied to the full backup are also stored, “jumping” in a backup chain behind the continuously updated full backup. This allows rolling back to the recent full backup in case you need to restore some older versions of your data.
The reverse incremental backup method is advantageous in terms of fast recovery, since it contains the most recent full backup file.
A smart backup is a combination of the full, incremental, and differential backups. Depending on the backup objectives and available storage space, the smart backup provides for efficient backup data and storage space management. The method follows a certain “smart” pattern to manipulate backup, cleaning, and merging operations. The table below gives an idea of how this backup type works.
|Source Data Action||Smart Backup Operations|
|1||Add 2 GB initial files||2 GB – Full|
|2||Modify 500 MB||500 MB – Incremental|
|3||Modify 500 MB||500 MB – Incremental|
|4||Modify 500 MB||500 MB – Differential|
|5||Modify 500 MB||500 MB – Incremental|
|6||Modify 500 MB||500 MB – Differential + backups # 2 & 3 are deleted|
|7||Modify 500 MB||500 MB – Incremental|
|8||Modify 500 MB||500 MB – Differential + backups # 4 & 5 are deleted|
|9||Modify 500 MB||500 MB – Differential + backups # 4 & 5 are deleted|
|10||Modify 500 MB||500 MB – Differential + backups # 6 & 7 are deleted|
By using the smart type of backup, you can always rest assured that you have enough storage for your backups with no risk of eventually running out of space.
Continuous Data Protection (CDP)
In contrast to other backups running on a periodic basis, the continuous data protection, sometimes called “continuous backup”, logs every change in the source data set, similar to the mirror backup. The difference is that in CDP the changes log can be rolled back to restore older states of data.
Synthetic Full Backup
A synthetic full backup has much in common with the reverse incremental backup type. The difference lies in the way of how the data is manipulated. The synthetic full backup starts from running a traditional full backup followed by a series of incremental backups. At a certain moment, incremental backups are consolidated and applied to the existing full backup to synthesize the most recent full backup as a new starting point.
The synthetic full backup type has all the advantages of regular full backups, while consuming less time and storage space.
Advantages of the synthetic full backup are:
- Fast backup and restore operations;
- Better storage management;
- Low storage space requirements;
- Low network workload.
If you want to learn more about the synthetic backup type, read our blog post "What is Synthetic Backup".
A forever-incremental backup differs from the usual incremental backup in terms of backup data organization and processing. Similar to all backup types described above, the forever-incremental backup starts with an initial full backup as a reference point to track changes. From that moment on, only incremental backups are made without any periodical full backups, hence its name.
To illustrate this, suppose that you made a full backup on Saturday. Starting the next day, incremental backups are made on a daily basis. On Sunday, two new blocks, A and B, are created in the source data set. On Monday, block A is deleted and a new block C is created in the source data set. On Tuesday, block B is deleted and a new block D is created in the source data set. The forever-incremental backup system tracks all daily changes, clearing off all duplicated data blocks to reduce the storage space in the backup repository. At the same time, appropriate references are added to recovery points indicating related data blocks and restore sequence.
Depending on the individual backup retention policy, after the creation of a series of incremental backups, the outdated backup data and recovery points are removed to free up storage space in the backup repository. All stored backup data is organized, so that both the initial full backup and retained increments enable full restore operation together.
Advantages of the forever-incremental backup are the same as those of the synthetic full backup:
- Fast backup and restore operations;
- Better storage management;
- Low storage space requirements;
- Low network workload.
On the one hand, there is no generally good or bad backup type. You have to consider what type is best for you to suit your specific organization needs dictated by your data protection policies, available storage, resources, and media, network bandwidth, service-level agreements, critical data areas, etc. On the other hand, forever-incremental and synthetic full backups have decisively modernized the backup process and are the most up-to-date data protection techniques, designed to meet the needs of most organizations.
Being a modern native VM backup solution, NAKIVO Backup & Replication uses the forever-incremental backup type to back up VMs and the so-called "full synthetic mode" to store VM backups in the backup repository. The product also offers a full package of outstanding technologies and features assembled in one solution, such as changed block tracking (CBT and RCT), data deduplication, data encryption and compression, skipping swap data, network acceleration, scheduling tools, and integration with remote storage or cloud services.
You are welcome to download the full-featured Free Trial of NAKIVO Backup & Replication or request a Live Demonstration below.