NAS Buying Guide: Redundancy Is Not Backup

Redundancy is not backup

A mirrored or parity-protected array keeps you running when a disk dies. That is its entire job. Delete a folder and it disappears from every disk simultaneously. Encrypt the array with ransomware and every copy is encrypted. Suffer a fire, a flood or a theft and the whole box goes at once.

So a NAS is best understood as a fast, capacious, always-available storage tier that itself needs backing up — ideally with versioned history and an off-site copy, exactly as described in our computer backup strategy. Most NAS operating systems include tools to push a copy to cloud storage or to a second unit elsewhere; configure that on day one.

Define the job first

These pull in different directions, which is why a general NAS buying guide recommendation is worth so much less than a specific one. Write down your job before comparing models, because a unit that is excellent as a backup target can be frustrating as a media server and vice versa.

Bays, capacity and the usable-space surprise

Redundancy costs capacity, and the arithmetic catches people out. Two 12TB drives in a mirror give you 12TB usable, not 24TB. Four 12TB drives with single-drive parity give roughly 36TB usable. You are buying insurance with disk space, and it is not a small premium.

Buy more bays than you currently need. Expanding an array later is possible on most systems but slow and constrained, whereas adding a drive to an empty bay is trivial. A four-bay unit populated with two drives is usually a better purchase than a full two-bay one at similar cost.

Use drives rated for continuous operation in multi-drive enclosures. Desktop drives work until vibration and duty cycle catch up with them, and the price difference is small against the cost of a rebuild. Drive selection is the cheapest insurance in any NAS buying guide.

Rebuild risk grows with drive size

When a drive fails in a parity array, replacing it triggers a rebuild that reads every remaining drive in full, at length, under sustained load. Two consequences follow, and they are the reason storage people are cautious.

First, rebuilds on large modern drives take a long time — frequently more than a day — and the array is degraded throughout. Second, that sustained read is precisely the workload most likely to surface a latent fault on a second drive, which is when a single-parity array becomes a total loss.

The mitigations: use double parity on arrays with many large drives, run scheduled integrity checks so latent errors surface before a failure rather than during a rebuild, and keep the off-site backup current so a failed rebuild is an inconvenience rather than a catastrophe. No NAS buying guide should treat rebuild risk as theoretical; it is the moment arrays are actually lost.

Processor, memory and transcoding

For a pure backup target, the processor barely matters. For media, it decides whether the unit can convert a video stream on the fly for a device that cannot play the original — and hardware-accelerated transcoding is the specific capability to look for, since it does the job at a fraction of the power a general-purpose processor needs.

Memory matters for file system caching, snapshots and anything you run in containers. Check whether it is upgradeable; many entry units solder it. Error-correcting memory is worth having on an archive you intend to keep for decades, and is generally found on higher tiers.

The network is often the ceiling

A NAS full of fast drives connected by gigabit Ethernet delivers roughly 110 MB/s and no more, which is slower than a single modern hard drive. If you move large files regularly, multi-gigabit networking on both the NAS and the client is the upgrade that changes your experience — more than drive choice does.

Wire the NAS, always. And confirm your router or switch has a port that matches, because a 2.5-gigabit NAS plugged into a gigabit switch runs at gigabit.

The checklist

• Configure an off-site backup of the NAS on day one.
• Write down the job before comparing models.
• Calculate usable capacity after redundancy, not raw capacity.
• Buy more bays than you need now.
• Use drives rated for continuous multi-drive operation.
• Schedule integrity checks and consider double parity on large arrays.
• Match network speed end to end — NAS, cable and switch.

Followed in that order, a NAS buying guide produces a machine sized for the work rather than for the specification sheet, and — more importantly — one that is not quietly serving as the only copy of everything you own.