Backup methods are essential for protecting your data from loss, corruption, or disaster. But not all backup methods are created equal. Depending on your business needs and goals, you may prefer one method over another. In this article, we will explain the advantages and disadvantages of different backup methods in terms of RPO and RTO.
RPO and RTO are two key metrics that measure the effectiveness of your backup strategy. RPO stands for recovery point objective, and it refers to the maximum amount of data that you can afford to lose in a backup event. For example, if your RPO is one hour, it means that you can restore your data to a point that is no more than one hour old. RTO stands for recovery time objective, and it refers to the maximum amount of time that it takes to restore your data and resume normal operations. For example, if your RTO is four hours, it means that you can recover your data and get back to business within four hours.
The balance between RPO and RTO is often a choice to make between the two. They are like levers that goes against each other. RTO is defined by the amount of "snapshots" or the frequency of backups you take, obviously, only differential or incremental backups are fast enough to sustain a high frequency of backups. RPO is the closest point to the latest snapshot, BUT, the more snapshots you take, the longer the RTO is, as restoring requires to rebuild through all the differential backups, which takes time to rebuild and verify.
RPO and RTO also have to be aligned with the needs of the business. Not all data necessarily should have the same RPO and RTO. Good backup and recovery systems will allow you to categorize data in such as way as to assign different policies governing RPO/RTO and aligning to business SLA's. In some cases, snapshots can be used entirely to meet those SLA's. In other cases, longer term backups should be used. In my experience, most customers require a combination of both.
A full backup is the simplest and most comprehensive backup method. It involves copying all the data from your source to your destination, without any filtering or compression. A full backup ensures that you have a complete and consistent copy of your data, which can be easily restored in case of a failure. However, a full backup also has some disadvantages. It requires a lot of storage space, bandwidth, and time to perform. It also has a high RPO, since it only captures the data at the time of the backup, and any changes made after that are not included. A full backup also has a low RTO, since it can take a long time to restore a large amount of data.
Once taking a full backup, the chosen solution must also consider an "application aware" function, or be scheduled in a way that allow proper recovery. If you have multiple workloads, combining database storage and file storage, where the database keep reference of stored files, it is critical that the backup process happens in a way that the 2 workloads are in sync, not to lose integrity when recovering.
An incremental backup is a more efficient and flexible backup method. It involves copying only the data that has changed since the last backup, whether it was a full or an incremental backup. An incremental backup reduces the amount of storage space, bandwidth, and time needed to perform a backup. It also has a low RPO, since it captures the data more frequently and minimizes the data loss. However, an incremental backup also has some disadvantages. It requires more complexity and management to keep track of the backup chain and dependencies. It also has a high RTO, since it can take a long time to restore multiple backup files and apply them in the correct order.
A differential backup is a hybrid backup method that combines some aspects of full and incremental backups. It involves copying only the data that has changed since the last full backup, regardless of any incremental backups. A differential backup offers some advantages over both full and incremental backups. It reduces the amount of storage space, bandwidth, and time needed to perform a backup compared to a full backup. It also simplifies the restoration process and reduces the RTO compared to an incremental backup. However, a differential backup also has some disadvantages. It requires more storage space, bandwidth, and time than an incremental backup, since it copies more data each time. It also has a higher RPO than an incremental backup, since it captures the data less frequently and increases the data loss.
A synthetic backup is an advanced backup method that uses a combination of full, incremental, and differential backups to create a synthetic full backup. A synthetic backup involves creating a full backup once, and then updating it with incremental or differential backups periodically. A synthetic backup has many advantages over other backup methods. It optimizes the storage space, bandwidth, and time needed to perform a backup, since it does not require copying all the data each time. It also improves the RPO and RTO, since it provides a consistent and up-to-date copy of the data that can be quickly restored. However, a synthetic backup also has some disadvantages. It requires more processing power and software to create and maintain the synthetic full backup. It also depends on the availability and integrity of the previous backups, which can be corrupted or lost.
When using synthetic backup, it is absolutely critical to have a complete backup validation mechanism by implementing full restore tests. By its nature, synthetic backup involves all the risks of the other methodologies (full, incremental and differential). Depending of the integrity check, the synthetic backup could be corrupted by a single error in a differential or incremental, leading to unsolvable situation.
As recommended in best practices, it is critical to test the backups doing a full restoration on a regular basis, backup failure are only identified at the time the recovery is needed otherwise. Backups should be also isolated from the network and use independent local credentials. In case a threat actor reach the infrastructure, a corrupted network should not allow access to the backup infrastructure. It is also critical to have offsite backups, encrypted ideally. One of the storage should be an immutable backup, where even admins can't delete the data. Typical ransomware attacks involves destroying the backups to make sure target is forced to pay the ransom to discover. Immutable backups are an efficient solution against this threat.