Set the usable capacity and the number of drive failures you can tolerate first, then model the drive layout, check any hardware you already own, pick the storage software, and estimate power cost before buying a box. Four 8TB drives give 24TB usable in RAID 5 or RAIDZ1, or 16TB in RAID 6, RAIDZ2 or RAID 10.
Contents
Provenance
- Oct 3, 2026Published
- Nov 18, 2026Last substantive update
Plan the data before the enclosure
Define usable capacity and growth, failure tolerance and backup, plus power assumptions before choosing a NAS or drive model. The project template opens a four-bay example with four 8TB WD Red Plus drives and RAID 5 or RAIDZ1 calculator state. Every one of those values is a planning default.
Start a NAS capacity and power project. The template opens the NAS and RAID Storage Calculator at four 8TB drives in RAID 5 and carries that layout into the Power Cost Calculator. It names the WD Red Plus 8TB (WD80EFPX) so the example has an exact drive identity, not because you own it or because it’s in stock; replace the drive, layout, electricity rate and wattage with your own before treating the example as a shopping list.
A NAS plan can look complete while four questions remain mixed together: how much data fits, how many drive failures the array survives, whether existing hardware is worth reusing, and what the finished machine costs to leave on. Answer them in that order. Product selection comes after the workload and failure plan.
1. Inventory the data and its growth
Start with the data you have now. Separate it into three groups:
- Primary data: files for which the NAS will be the working copy.
- Replicas: data already held somewhere else and safe to recreate.
- Replaceable data: downloads, caches, media transcodes, and other material that can be rebuilt.
Record the current amount for each group, the expected growth, and how long the system should last before you expand it. A single total hides the difference between ten terabytes of family photos and ten terabytes of replaceable media.
Then set the usable-capacity target: the space left after the RAID or RAIDZ layout, before snapshots, filesystem overhead and free-space headroom. Keep those reservations visible instead of treating every reported terabyte as available.
I would also name the largest single restore. Restoring one workstation, one media library, or one project archive can change the network and backup plan even when total capacity stays the same.
2. Choose the failure tolerance before the drive count
Write down how many drives can fail at once without losing the array, and what happens while a failed drive is replaced. Failure tolerance is the requirement; the RAID label is only how you get it.
For the template’s four 8TB drives, the NAS and RAID Storage Calculator works out:
| Layout | Usable space | Drive failures survived |
|---|---|---|
| RAID 5 / RAIDZ1 | 24TB | 1 |
| RAID 6 / RAIDZ2 | 16TB | 2 |
| RAID 10 | 16TB | 1 guaranteed (2 only if they’re in different mirror pairs) |
Those figures are before filesystem overhead and free-space headroom. Test the smallest drive count that meets your target, then one layout with more fault tolerance. The calculator does the capacity math; it can’t decide how much downtime or rebuild exposure your data can take.
For a large six-bay plan, read Six-Bay NAS Drives: The Rebuild Math at 20TB+, which ties its layout and rebuild conclusions to a named drive size and dated sources. For the drives themselves, best NAS hard drives compares models and cost per terabyte.
RAID and RAIDZ keep a service running through some drive failures. They do not create an independent backup. Record where the second copy lives, how it’s protected from the primary system, and when you last tested a restore.
3. Qualify the hardware you already own
If an old desktop is available, don’t price a new NAS until it has passed the basic checks in the old-PC NAS guide: CPU, memory, SATA ports, drive bays, a separate boot device, the network path, HBA use and operating-system choice. If you’re weighing a ready-made NAS instead, Synology or build runs the five-year cost.
Capture the exact machine in the project:
- CPU and motherboard model.
- Installed memory and available slots.
- Native SATA and M.2 ports, plus occupied ports.
- Physical bays and airflow across the intended drives.
- Network interface and the slowest link to the main client.
- Existing power supply, connectors, and age.
- The boot device and where configuration backups will live.
Mark anything you haven’t opened or measured as unknown. “Old gaming PC” is not a hardware record; the motherboard model and a photo of the drive cage answer questions a processor name can’t.
Use the AM5 ECC Compatibility Checker only when the plan uses an AM5 platform it covers. A missing board or an unverified firmware path stays a limitation; don’t treat a nearby model’s result as evidence for yours.
4. Pick the storage software from the operating constraint
The operating system follows the hardware, drive layout and who maintains it. The Unraid vs TrueNAS vs OMV vs Proxmox comparison covers the choice for a storage box, and the storage backend comparison covers storage under a hypervisor. Keep three questions visible:
- Will this box primarily serve storage, run applications, or host virtual machines?
- Are the drives matched, or does the project need to mix capacities?
- Who will maintain updates, replacement procedures, alerts, and backups?
Record the chosen software and the reason in the project. “Popular” is not an operating requirement.
Don’t install onto the only copy of any data. The project should show a separate temporary location or backup for every disk that will be erased during setup.
5. Carry capacity into power, then replace assumptions
Once the drive count and platform are plausible, continue from the NAS calculator to the Power Cost Calculator. The handoff carries the drive layout, but capacity doesn’t set wattage: the calculator starts from a labeled estimate that you replace with your hardware’s idle and load draw. 24/7 Homelab Idle Power Costs shows which of its rows are measurements and which are estimates.
Enter your electricity rate and expected load schedule. When the hardware exists, measure idle at the wall after services settle and measure the workload you care about, then replace the estimate and note the meter and date. Power can decide whether free old hardware stays the cheaper plan, or show that replacing a working machine is hard to justify.
6. Finish with a recovery drill
The project is ready for implementation when it names:
- Current data, growth window, and usable-capacity target.
- Chosen fault tolerance and a second-copy location.
- Drive count, size assumption, and an alternate layout.
- Exact existing hardware or the unresolved product categories.
- Storage software and the job it serves.
- Estimated idle/load power, electricity rate, and a plan to measure the result.
- One replacement procedure and one restore test.
After setup, unplug nothing at random. Use the platform’s documented failure procedure or a non-destructive test environment, then record what the restore or replacement drill actually proved. If a fault appears during setup, open the Diagnostic Workbench and keep the symptom, recent change and tests with this project. Export the project once the NAS reaches a known-good state.
Updates and corrections
- Last updated
- Methodology
- See our methodology for research and review standards.
- Update log
- 2026-11-18 — Added a usable-capacity table for the four-drive example, links to the NAS OS, drive and buy-versus-build guides, and cut the tool-operation detail.
- Corrections
- Spotted an error? Email a correction with the affected line and what you found. The draft includes this page’s address. Verified changes are recorded here. Contact details.