Losing data hurts, and it usually happens at the worst possible time.Just when you need those family photos, your final year project, the quarterly accounts, or the project you're due tomorrow. A hard drive crashes, ransomware encrypts everything, an accidental deletion… and the only question that matters is whether you had a good backup.
The key is not just making backups, but choosing the right method and support.Protecting a home laptop is not the same as protecting a data center with hundreds of virtual machines, nor does it make sense for a micro-enterprise and a multinational to use the same strategy. There are different types of copies (full, incremental, differential, mirror, synthetic, forever incremental, CDP…) and many use cases, plus a good handful of very powerful commercial solutions.
Classic backup types: full, differential, and incremental
The foundation of any data protection strategy revolves around three traditional copy typesThe basic backups are full backups, differential backups, and incremental backups. Almost all modern backup software variants are built upon these basics.
Full backup
Advantages of a full copy:
- Very quick and easy recoveryYou only need the latest complete copy (or the one corresponding to the desired point in time).
- Simple managementThere's no need to "rebuild chains" of copies; restoring is basically copying back.
- Fewer dependenciesThe integrity of a single file or set of files ensures restoration.
Major disadvantages:
- Brutal consumption of spaceEach full backup stores all the data, even if almost nothing has changed.
- Very long backup windowsIt can take up to ten times longer than other methods, saturating the network and disks.
- Impact on productionIf not programmed properly, the full backup will notice a difference in system performance.
It is usual to use the full copy as a starting point of a backup plan: an initial full backup is made (and then periodically, for example weekly or monthly), and differential and incremental backups are based on it for day-to-day use.
Differential backup
A differential copy saves all changes that have occurred since the last full copy.In other words, each new differential always "looks" at the same anchor: the last full.
Advantages of differential copying:
- Faster than a fullbecause only modified or new files are backed up since the last full backup.
- Relatively agile restoration: the last full and the last differential are sufficient.
- Fewer dependencies than incrementalIf one differential fails, a subsequent one is usually still valid.
Drawbacks to watch out for:
- It grows over timeEach day includes all the changes accumulated since the full backup, so Thursday's backup is usually larger than Monday's.
- Medium-high storage usageIt consumes less space than making full daily backups, but considerably more than a purely incremental scheme.
Differential backup is a good fit when you're looking for a balance between restoration speed and space consumption.especially in companies that can afford to take on more storage in exchange for faster recovery after an incident.
Incremental backup
Key advantages of incremental:
- Very fast backups: little data is copied if the interval between copies is short.
- Significant storage savingsThere are no duplicates of blocks or files that are already in the repository.
- Ideal for frequent programmingCopies can be run every hour or even more often without overloading the system.
Disadvantages and risks:
- Slower and more complex restorationsTo recover a specific point, you need the initial full and all subsequent incrementals up to that point.
- Fragile chainIf a piece of the chain is corrupted or lost, you could lose an entire section of historical data.
In practice, many advanced backup solutions mitigate these drawbacks. Reconstructing complete images "on the fly", using deduplication and intelligently managing the chain to speed up restorations.
Quick comparison between full, differential and incremental
If we look at the three classic types at a glance, the pattern is clearWhat you save in time and space when making a backup, you pay for (partially) when restoring.
| Criterion | Full Equipe | Differential | Incremental |
| Occupied space | Very high | Medium/high | Low |
| Copy speed | Low | Media | High |
| restore speed | Very high | High | Slower |
| What do I need to restore | Only the selected full | Last full + last differential | Full initial + all incremental |
| Ease of use | Very simple | Media | Media |
Advanced methods: mirror, inverse incremental, smart, CDP, fully synthetic, and forever incremental
Around the three basic types, variants have emerged designed to reduce the impact on production.to improve recovery times (RTO) and shorten the maximum acceptable data loss (RPO). Let's look at the most important ones.
Mirror backup
The mirror backup maintains an identical replica of the source.Same folder structure, same files, and, most importantly, only the latest version of each item. It doesn't create a history, but rather a "live clone".
Positive aspects:
- Direct access to files from the file explorerwithout needing to restore an image.
- Very fast recovery after a hardware failureSimply use the mirror copy as an immediate replacement.
- Operational simplicityThe process is usually similar to synchronization.
Weak points that make it delicate:
- It does not protect against deletion or corruption at the source.If you delete a file, it also disappears from the mirror.
- No previous versionsYou cannot go "back in time" because you only retain the last state.
- Higher risk of unauthorized access if it is not encrypted, since the files are "in plain sight".
The mirror is useful as an extra layer for high availability. (for example, replicating a file server), but it needs to be accompanied by another strategy with historical data (full+incremental, CDP, etc.).
Reverse incremental
The reverse incremental backup starts with a full backup, and from there, each new copy updates that full backup. so that it always reflects the most recent state, saving changes "backwards" in time in separate incrementals.
What is achieved with this approach?:
- Always have a very recent full backup ready to restore., with a minimum RTO.
- Preserving historical data thanks to the incremental data contained in ancient deltas.
- Greater chain resilienceIf an old incremental fails, the last recovery point (the synthesized full recovery) remains usable.
It is typically used when the absolute priority is to restore the latest version of the data as quickly as possible.For example, in business-critical systems where every minute of downtime translates into clear losses.
Intelligent backup (full + incremental + differential combination)
Some solutions implement what they call intelligent backup: an engine that decides whether to make an incremental copy, a differential copy, or even consolidate previous backups, depending on the volume of changes and retention policies.
Typical example of operation:
- An initial full is performed when the first data is added.
- When there are small changes, they are made incrementally.
- Every so often, the system generates a differential, consolidating and even eliminating old incrementals to save space.
The advantage is that you get multiple restore points without having to manually design all the cleaning and consolidation cycles; the software optimizes the space and backup windows automatically.
Continuous Data Protection (CDP or continuous backup)
Continuous protection records every change in near real time.Every time a file is modified, the system captures that change and sends it to the repository, allowing you to revert to almost any point in time.
What does CDP offer compared to regular copies?:
- RPO practically zeroThe amount of data you can lose is minimal, ideal for databases or financial systems.
- Recovery to very fine points over timeuseful against logical corruption, human error, or malware.
- It resembles a reversible change lognot just individual photos.
However, it does require infrastructure and design.: increased network and storage load, the need to protect the CDP system itself, and clear retention policies to prevent the repository from growing uncontrollably.
Full synthetic (synthetic full)
A full synthetic copy does not reread all the data from the source server.Instead, it is built in the repository by combining the initial full with the accumulated incrementals.
Benefits it brings:
- Reduce backup time in production: there is no need to go through the entire source system again.
- Less impact on network and CPU of the servers that are protected.
- It offers complete and recent restore points with the same content as a traditional full version, but generated much more efficiently.
Many modern manufacturers have standardized the use of full synthetics as part of their "full weekly copy" policies without actually reading everything from scratch at the source.
Forever incremental backups
The forever incremental model starts with an initial full value and, from then on, only incremental changes are made.New full backups are not planned from scratch; what is done is to internally reorganize the backups to maintain efficient restoration.
Interesting things about this approach:
- It minimizes the recurring impact on the source infrastructure.Because you should never repeat a full session.
- It combines very well with deduplication and compression, achieving compact repositories.
- The tools manage references and recovery points so that outdated historical data can be removed without breaking consistency.
It is common for solutions that offer forever incremental to combine it with fully synthetic ones. and scheduled consolidations, thus achieving fast copy times and agile restorations.
Comparison of speed, space, and cost between incremental and differential
When delving into real-world strategies, the question almost always arises of whether to focus more on incremental or differential growth.Both are "smart" methods that save time and disk space compared to full methods, but they behave differently in the medium term.
In terms of copy speedIncremental backups usually win out: by copying only what has changed since the last backup (of any type), the data volume is typically very low if backups are frequent. Differential backups, by accumulating all changes since the last full backup, tend to become more resource-intensive the further back in time that full backup is.
If we look at storage spaceThe picture is similar: differential changes take up more space because they store larger sets of changes, while incremental changes break those changes down into daily (or hourly) chunks, resulting in many small but, overall, more compact files.
In pure and simple terms, cost-cutting terms (Especially in cloud services where you pay per GB and per operation) incremental storage has an advantage in the short term, but if historical data is stored for months or years and retention policies aren't managed properly, you can end up with very long chains that also consume space. That's where automatic cleanup, consolidation, and global deduplication come into play.
In terms of restoration speed, the differential outpaces the incremental speed.You need fewer pieces (full + last differential) compared to the full stack of incrementals starting from the last full. That's why differential is usually recommended for companies with a high need for fast recovery, and incremental for those who prioritize optimizing storage.
Physical storage media and formats for your backups: HDD, SSD, NAS, USB and cloud
Just as important as the backup method is where you store those backups.It's not the same to use a small USB drive for home photos as it is to set up a redundant NAS for an SME or a complete BaaS in the cloud for a corporation.
External hard drives (HDD and SSD)
Mechanical hard disk drives (HDDs) remain the kings of cheap storageThe cost per GB is much lower than that of SSDs, making them ideal for storing large volumes of historical backups.
Key aspects to consider when choosing an HDD for backups:
- 2,5″ vs 3,5″ format2,5″ drives are more portable and are usually powered by USB; 3,5″ drives offer more capacity (8 TB, 10 TB or more) but require their own power supply.
- Interface and speedThese days, USB 3.0 or higher is the norm; check the actual average read/write speeds, not just the maximum theoretical speeds.
- Physical robustnessIf you're going to move it around, make sure it has a sturdy casing, a non-slip grip, and some shock resistance.
- Included softwareSome models come with their own backup and encryption tools, useful if you don't want to deal with third-party software.
External SSDs, on the other hand, shine in speed.Ideal for working on real-time projects or for very fast restorations, although at a significantly higher price per GB, and it's worth checking the SSD firmware.
NAS: your private cloud for backups
A NAS (Network Attached Storage) is, essentially, a mini network storage serverPerfect for centralizing backups across multiple computers and providing remote access to data, and it typically integrates with SMB file synchronization.
When evaluating a NAS for backups It is worth paying attention to:
- Hardware power (CPU and RAM)Although backup isn't the most demanding task in the world, a multi-core CPU and at least 512 MB – 2 GB of RAM helps to handle multiple tasks simultaneously (backups, streaming, remote access, etc.).
- Number of bays and maximum capacityThe more you think your data will grow, the more a 2, 4 or more bay NAS with RAID capability will benefit you.
- Operating system and appsMany brands integrate backup tools, synchronization with public clouds, replication between NAS, and snapshot functions that fit very well into backup strategies.
A well-configured NAS allows for the application of very robust strategies.Features include: fast local backups, replication to another NAS at a different location, integration with cloud services, near-instant snapshots… It is a very common piece of equipment in SMEs and advanced homes.
USB flash drives and specific solutions for mobile phones
USB flash drives aren't the most robust or fastest storage medium, but they still have their place. for partial copies or transferring data from one site to another.
For lightweight backups (documents, small collections, one-off exports) These may be sufficient, and some models include dual connectors (USB-A and USB-C or Lightning) and their own apps for making copies of photos and videos directly from mobile phones and tablets.
They shouldn't be your only line of defense in the face of serious disasters, but they do fit as a quick and cheap complement within a broader strategy (for example, to remove a monthly manual copy of critical data from the office).
The cloud and the Backup as a Service model
Cloud backup services (BaaS) allow you to automate backups to secure external data centers.for example, automating the copying in OneDrivewith encryption, advanced retention, and granular restoration options.
Clear advantages of the cloud approach:
- Integrated off-site copyThe 3-2-1 rule is easily met by having at least one copy in a different location.
- ScalabilityYou can expand space and functionality without buying new hardware.
- advanced functionality: file recovery, mailbox recovery, VM recovery, orchestrated DR testing, etc.
Generalist and specialized manufacturers and providers (AWS, Azure, Dell EMC, NetApp, Oracle, Sarenet, VMware, Acronis, NAKIVO, etc.) offer solutions for almost any scenarioFrom home offices to large hybrid data centers with on-premises and cloud-native workloads.
Choose the backup method according to the size and needs of the company
A micro-enterprise with five employees should not consider the same backup design as a company with hundreds of serversThe volume of data, the budget, the importance of the systems, and the legal obligations completely change the approach.
Micro-enterprises: simplicity, automation and the cloud
Very small businesses usually have few computers, some office software, perhaps a small cloud-based ERP, and a lot of personal data. (customers, invoices, emails). They are just as obligated as any large company to comply with the GDPR, but their resources are limited.
Recommended scheme:
- Full weekly copy of the key equipment (administrator's PC, accounting, etc.).
- Automatic daily increments during off-peak hours.
- Cloud backup with encryption, either attached to Microsoft 365/Google Workspace or through a simple BaaS.
- Monthly restoration trial to verify that the backups can actually be recovered.
The idea is that the person in charge doesn't have to be constantly monitoring or struggling with complex configurations.Everything must be programmed and monitored with simple alerts.
SMEs: efficiency, hybrid on-premises + cloud and the 3-2-1 rule
In a typical SME, we already find physical or virtual servers, NAS, ERP, CRM, mobile devices and often several locations.Here, data loss or prolonged downtime can result in direct financial losses.
Good practices:
- Full weekly + daily incremental (or differentials if rapid restoration is prioritized).
- Local NAS with RAID for fast backupsPlus cloud replication or to a second external NAS.
- Apply the 3-2-1 or 3-2-1-1 rule: three copies of the data, on two different types of media, one off-site and, optionally, one immutable or disconnected.
- Quarterly review of backup policies and documented recovery tests.
Here, intelligent schemes such as fully synthetic, inverse incremental, or forever incremental fit very well.because they reduce bandwidth and storage consumption while maintaining reasonable RPO/RTO.
Medium and large companies: business continuity and strict compliance
In organizations with multiple data centers, massive virtualization, Kubernetes, SaaS, and hybrid environments, backup is integrated into the business continuity and disaster recovery strategy.It's not just about saving files: it's about orchestrating the recovery of entire applications.
Common elements in this segment:
- Extensive use of CDP for critical systems with near zero RPO (databases, core banking, etc.).
- Full synthetic and incremental backups forever to reduce continued impact.
- Immutable copies (WORM, object-lock) and air-gapped to protect against advanced ransomware.
- DR plans tested periodically, with drills and RPO/RTO compliance reports.
The solutions here are usually unified data protection platforms (Dell EMC PowerProtect, VMware Cloud Disaster Recovery, Arcserve, NetApp, Oracle Object Storage with replication, etc.) with centralized management, global deduplication, monitoring and reporting for audits.
Regulatory compliance in Spain: GDPR, LOPDGDD, ENS and ISO 27001
Many companies think backups are a "good practice", but as soon as you handle personal data it becomes a legal obligationThe regulations require the ability to guarantee confidentiality, integrity, and availability.
Regulatory keys:
- GDPR and LOPDGDDThey require data controllers to be able to quickly recover personal data if there is a physical or technical incident.
- ENS (National Security Scheme): applicable to the public sector and many technology providers, it explicitly includes copies, replicas, off-site, immutability and periodic testing.
- ISO/IEC 27001 and 27002: reference standards in information security management, which include specific controls on backups, retention, and restoration testing.
It's not enough to say "we make copies"The backup policy must be documented, defining responsibilities, retention cycles, methods used, encryption, backup locations, and the frequency of recovery tests. If an incident occurs and the Spanish Data Protection Agency (AEPD) determines that personal data could not be recovered due to a lack of backups, the penalties can be substantial.
Understand the differences between full, incremental, differential, mirror, CDP, and forever incremental backup schemes It allows you to design a data protection strategy that aligns with the real needs of each environment, optimizing resources without jeopardizing business continuity or legal compliance. Those who take backups seriously sleep more soundly… and when something does break, they notice it much less.
