What a RAID does – and what it does not
The concept is quickly explained: several hard drives or SSDs – connected via SATA or as NVMe drives – work together as one logical drive. Depending on the RAID level, data is mirrored (RAID 1), distributed across several disks (RAID 0) or secured with parity information (RAID 5, RAID 6). If one disk fails in a redundant array, the system keeps running; the controller then reconstructs the missing portion from the redundancy – a process known as a rebuild.
A RAID therefore serves exactly one purpose: fault tolerance and availability. Operations continue despite a hard drive failure, without applications stalling or a server having to be rebuilt. Configurations such as RAID 10 additionally increase throughput and performance when accessing large volumes of data – for example on a workstation handling video footage, CAD data or AI models, or on a virtualisation host whose VMs share the storage resources.
What a RAID does not provide is a time-shifted, independent copy of your files. That is precisely what a backup is – the overview shows the comparison:
| Risk | RAID (redundancy) | Backup (data backup) |
|---|---|---|
| Failure of a single disk | covered | covered |
| Accidentally deleted files or folders | open – the deletion is applied immediately | covered |
| Ransomware, malware and viruses | open – the encryption affects the entire volume | covered if the backup is kept separate |
| Defect in the controller, mainboard or power supply | open | covered |
| File system error, faulty update, software fault | open | covered |
| Fire, water damage or theft at the location | open | covered with an off-site copy |
| Second disk failure during the rebuild | open | covered |
Redundancy works in real time, a backup does not
Timing is the decisive factor. A RAID works in real time: every change lands on all participating drives at the same moment. That is by design, but it makes the array blind to the cause of a change. Whether a file was saved deliberately, overwritten by mistake or encrypted by ransomware, the controller does not distinguish. A backup, by contrast, creates a state from yesterday, from last week or from last month – and that gap in time is its actual value.
Four misconceptions come up particularly often in practice:
- "The NAS runs RAID 5, so the data is safe." The array protects availability, not content. One wrong click is enough, and the files disappear from every disk at once.
- "The cloud folder syncs everything anyway." Synchronisation is mirroring, not a backup. Without versioning, an encryption trojan on the PC also renders the copy at the provider unusable.
- "A rebuild will work out." Anyone who has had to wait through the rebuild of a large array knows the strain involved: every remaining drive is read in full over hours or days. With HDDs holding several TB of capacity, this increases the likelihood that previously unnoticed bad sectors will surface at exactly that moment – taking the array with them.
- "A second disk in the same enclosure is enough." Same power supply, same location, same risk: in the event of a power surge, fire or theft, both storage devices are affected.
The 3-2-1 rule as the framework for your backup strategy
One concept has become established for data backup that requires no special infrastructure and works just as well in a company as on a private home server:
- 3 copies of the important data – the original on the production system plus two backups.
- 2 different media: the internal storage volume plus an external hard drive on the USB port, a second NAS on the network, a tape drive or cloud storage.
- 1 copy off site at a different physical location – kept at home, at a second company site, with a cloud service or on a rented server or VPS.
Data loss despite RAID? We assess the condition of your array.
If an array no longer mounts, a rebuild has aborted or several disks have failed at once, the system should be left untouched – every further start-up attempt can change the situation. For an initial technical assessment, reach us on 0800-881 12 25 (24 h) or via the contact form.
Building RAID sensibly into your data backup
The two approaches are not mutually exclusive: they pursue different goals, and the advantages of one offset the disadvantages of the other.
What the RAID handles
- Keeping operations running during a hard drive failure
- Minimising downtime for servers, virtual machines and applications
- Improving read and write speed for large volumes
- Bundling the storage space of several disks into one drive
What the backup handles
- Retaining earlier versions of documents, images and project data
- Restoring accidentally deleted or overwritten content
- Returning to a clean state after a malware infection
- Securing the data independently of device and location
In practical use, this means the RAID is part of the storage architecture, not of the backup plan. When buying a NAS, it is worth planning the backup target straight away. A setup with four bays and hard drives designed for continuous operation – such as the WD Red series – is a good place for the central storage of your data, as long as that array is regularly backed up to another device; the same principle applies to an individual computer. What else such a system can be used for is described in our article on using a NAS as a cloud server.
Snapshots, checksums and automated backups
Modern systems offer functions that go well beyond pure redundancy. Anyone setting up a NAS or home server should know these methods and use them:
- Snapshots at file system level preserve earlier states of a volume without doubling the storage space. ZFS and Btrfs handle this, as does the management software of widely used NAS platforms from Synology and QNAP or distributions such as TrueNAS.
- Checksums detect silent bit errors and report damaged blocks before they find their way into the backup unnoticed. Combined with ECC RAM – depending on the mainboard and chipset, for instance on Intel platforms – data integrity increases further.
- Incremental backups transfer only the changes since the last run. On Linux, rsync is the classic tool; on Windows, the operating system or a dedicated backup program takes on this task.
- Snapshots of virtual machines are explicitly not part of this: they sit on the same datastore as the VM and fail along with it, as our page on data recovery for virtual systems shows.
- Replication and monitoring copy snapshots automatically to a second device and report faults by e-mail or app – so you notice a problem before it turns into an outage.
One point is easily overlooked: a backup only counts as a backup once the restore is demonstrably working. Test in quiet periods whether individual files and, in an emergency, a complete system can be restored. The test costs little time and money and is the most reliable indication of the state of your data security: experience shows it quickly reveals whether daily use and the backup plan fit together – and you know what to expect. Further measures are collected in our tips on preventing data loss and in the blog.
When the array fails anyway
Even a well-planned configuration can reach its limits: several defective drives, a lost RAID configuration after a controller replacement, an aborted rebuild or a damaged file system. How to recognise such cases early is described in our overview of the symptoms of a defective RAID.
The right response is then to avoid putting further strain on the system: no repeated rebuild attempts, no reinitialisation of the array, no recovery software writing to the original disks. Each of these steps can make reconstructing the original structure more difficult. It makes sense to switch the device off and agree the next steps with a specialist – the first measures in the event of data loss summarise this concisely.
We have been reconstructing failed storage systems in our own laboratory since 2004 and are an experienced partner for companies and private users alike. Professional RAID data recovery involves deriving the order and parameters of the array from the individual disks and reassembling the data from verified images; with a NAS, its own volume structures are added to that. We describe the data recovery process step by step.
Conclusion: redundancy and backup belong together
A RAID is a sound choice when systems need to stay reachable despite a disk defect. It does not replace data backup, because it applies every change – including every error – to all drives immediately. Combining both covers two different risks: the array protects the hardware, the backup protects the data. Three copies, two media, one of them off site, plus versioning and a regularly tested restore – that puts your infrastructure in shape for an emergency.
Frequently asked questions about RAID and backups
Is RAID 1 sufficient as a data backup?
No. RAID 1 mirrors two hard drives in real time and keeps operations running if one of them fails. Everything else is mirrored unchanged: deleted files, data encrypted after a malware infection, or a damaged file system. A defect in the controller or damage at the location also affects both disks at once. RAID 1 is therefore an availability solution and should always be complemented by an independent backup.
How often should I back up the data from a NAS?
That depends on how much work you would be willing to repeat in an emergency. If content changes daily, an automated daily run makes sense; for a private archive, a weekly or monthly backup is often enough. Regularity matters more than frequency: an automatically executed job with logging and notification is more reliable than a manual copy that nobody remembers to make.
Can data be recovered from a failed RAID without a backup?
In many cases, yes. Even without a backup, the user data usually remains on the individual disks; the decisive part is reconstructing the original configuration of the array correctly. How well the recovery succeeds depends on the condition of the storage devices and on which write operations took place after the failure. A reliable assessment is only possible after analysing the individual drives.
Lars Müller