3-2-1 Backup Sizing Calculator for a NAS

The array is copy one. That is the whole reason this calculator exists, because almost every home server budget covers the enclosure and the drives and then quietly stops, leaving parity doing a job it was never able to do. RAID rebuilds a failed drive and protects against nothing else, so copies two and three are not optional extras, they are the part of the build that survives a deletion, a ransomware run, a fire or a theft. This tool sizes them and prices them.

How much backup capacity does a home server actually need?

Size the backup against the data you hold, not against the capacity of the array. This single reframing is where most of the money is saved. An eight-bay array running RAID 6 with eight 16 TB drives has 96 TB usable, and the household filling it might be holding 11 TB of actual files. Buying 96 TB of backup target to match the array would cost thousands of dollars to protect data that fits on one external drive.

The sizing formula has four terms and the calculator above applies all of them:

1. Used space today. Not allocated, not provisioned, used.
2. Growth to your horizon. Compounding, because libraries grow by percentage rather than by a fixed number of terabytes a year.
3. Version history overhead. Keeping yesterday's copy of a file costs space, and that is the term that makes a backup useful against ransomware rather than only against hardware failure.
4. Headroom. A backup target running at 99 percent is a backup that fails silently the next time something large lands.

Three years is the right horizon for most people. Beyond that, drive prices per terabyte and available capacities both move enough that buying ahead is usually worse value than buying again later, and the money spent early is capacity sitting idle.

What exactly is the 3-2-1 rule?

Three copies of the data, on two different kinds of media, with one copy off site. It is a rule about independent failure modes rather than about redundancy, which is precisely why an array cannot satisfy more than one part of it.

Count carefully. The array is copy one. An external drive plugged into it is copy two, and it is on different media in the sense that matters, which is that it is a separate device with a separate controller and it can be unplugged. A drive kept at a relative's house, or a cloud bucket, is copy three. What does not count is a second volume on the same NAS, a second partition on the same drives, or a snapshot, because all three die with the chassis.

The most common mistake is calling the array two copies because it has parity. A RAID 6 array survives two drives dying. It does not survive somebody deleting a folder, a ransomware process encrypting every share it can reach, a power supply failure writing garbage across all members, a house fire or a burglary. Parity and backup protect against disjoint sets of events, and you need both. The full argument is in the 3-2-1 backup rule guide.

What does version history really cost in space?

Version history is the extra space consumed by keeping previous states of files that have changed. Modern backup tools store versions as deltas or as deduplicated blocks, so the cost is driven by how much of your data actually changes rather than by how many versions you keep.

That is why the dropdown in the calculator is phrased by content type. A 4K film library is written once and never edited, so a year of retention adds almost nothing. A folder of active project files, virtual machine images or databases rewrites large blocks constantly and can double the space a single current copy needs.

Version overhead is the extra capacity above one current copy of the data, assuming a backup tool that stores changed blocks rather than whole files. Whole-file versioning, which some simple sync tools use, costs dramatically more and is worth avoiding on large data sets.

Which backup target is cheapest per terabyte?

Backup capacity is the one place in a home server build where you are buying bulk storage with no performance requirement at all, so cost per terabyte is close to the only number that matters. The table below prices every verified target against that measure.

Prices are the listing prices recorded when each product was verified and they move often, so confirm on Amazon. Cost per TB uses decimal terabytes as sold. The cheapest target here is the Seagate Expansion Desktop 16TB External Hard Drive at $30.09 per terabyte.

Two patterns fall out of that table and they hold across price changes. Large external desktop drives beat both small externals and bare drives per terabyte, because you are buying one enclosure and one power supply amortised over more capacity. And portable bus-powered drives cost substantially more per terabyte, which is a fair price for the offsite copy specifically, because a drive that fits in a coat pocket is a drive that actually leaves the house.

Should copy three be a drive or the cloud?

Cloud storage is an excellent copy three and a poor copy one. It is off site by definition, it survives events that reach your whole house, and it requires no discipline from you once configured. The two costs are money at scale and time on restore.

Run the restore arithmetic before deciding, because it is the part people skip. Recovering 10 TB over a 200 Mbps download link takes about 4.6 days of continuous transfer at wire speed, and considerably longer in practice. The same 10 TB read from a local external drive at 180 MB per second takes about 15 hours. That difference is why the fast copy should be local and the slow copy should be the one you hope never to need.

The money crossover is worth checking too. At typical consumer cloud pricing, tens of terabytes cost more per year than the drives would cost outright, and the drives keep working after year two. Where cloud wins outright is small, high-value data sets: documents, tax records, code and photographs, where the total is a few hundred gigabytes and the value per byte is enormous. Many households are best served by both, with the cloud holding the irreplaceable core and a rotated drive holding the bulk media.

How should the offsite copy actually be rotated?

An offsite drive is only as current as the last time it came home. The working pattern is two identical drives: one connected and updating, one at a second location. Swap them on a schedule you will genuinely keep, which for most people means monthly rather than weekly.

Three details make the difference between a rotation that works and one that quietly stops:

• Encrypt the drive that leaves the house. A backup drive is your entire life in a bag. Full-disk encryption on the target costs nothing and removes the worst case of losing it.
• Never leave the offsite drive permanently mounted when it is home. A backup share reachable from an infected machine gets encrypted along with everything else, which is the failure mode a third copy exists to prevent.
• Label the drives and record the swap. Two identical black boxes with no markings is how people end up overwriting the only good copy.

A dual bay drive dock makes the rotation cheaper still, because it lets you use bare drives as targets and swap them like cartridges instead of buying an enclosure with every drive. The same dock is the tool you would already own for burn-in testing new drives before they go into the array.

What about the copy the calculator cannot size?

Two things sit outside the arithmetic and are worth naming before you buy anything.

The first is restore testing. An untested backup is a hypothesis. Once a quarter, restore a real folder to a different location and open the files. This finds the problems that no capacity figure predicts: a job that has been failing silently for months, an exclusion rule that quietly skipped the one directory that mattered, an encryption passphrase nobody wrote down.

The second is what you have decided not to protect. Most households have a tier of data that genuinely does not need three copies, usually a re-downloadable media library, and a much smaller tier that is irreplaceable. Splitting them explicitly lets you put the irreplaceable tier in the cloud cheaply and keep the bulk tier on local drives, and it usually cuts the number the calculator returns by more than any other decision.

Related reading

• The 3-2-1 backup rule guide for the full method behind this tool
• Best external drives for NAS backup for the three-tier picks
• NAS against cloud storage for the copy-three decision in detail
• NAS security hardening because ransomware is the threat a third copy answers
• RAID capacity calculator to size copy one across every layout
• Cost per terabyte chart for the current shape of drive pricing