Leo · 2026-10
This small office NAS backup window planning test uses three inputs only: published woCyber hardware specs, arithmetic you can redo, and one third-party figure with a date.
The hardware basis is the X4 (N100, 4 bays, 120 TB, 2 × 2.5GbE) and the S2 (2 bays, 60 TB, 2.5GbE). Both are export SKUs, not retail units.
We hold link efficiency at 0.85 across every calculation. That covers TCP overhead, SMB chatter, and the fact that no real transfer hits line rate.
Sample size is one job shape: a 6 TB incremental backup set.
Time window is a single overnight run. We did not test spinning-disk seek behaviour or mixed read/write loads.
Where a number cannot be traced to a spec sheet or a formula, it is stated as an assumption rather than a measurement.
If the bottleneck in a small-office NAS backup is the network link rather than the drives, then moving from 1GbE to 2.5GbE should cut the window by roughly 60% on the same disks.
That is falsifiable. If the drives were the constraint, the two runs would land within 15% of each other.
Start with the formula, because you will want to run it on your own numbers. Any small office NAS backup window planning test lives or dies on this one equation.
Backup window in hours = backup set (TB) × 8 × 1024³ ÷ (link speed in GbE × 10⁹ × 3600 × efficiency).
Plug in 6 TB and 1GbE at 0.85 efficiency: 6 × 8 × 1,073,741,824 = 51,539,607,552 bits.
Divide by 1 × 10⁹ × 3600 × 0.85 = 3,060,000,000,000. You get 0.0168 × 1000, or 14.9 hours.
Same job on 2.5GbE: the denominator becomes 7,650,000,000,000. The result is 5.9 hours.
[CHART: Backup window for a 6 TB set, 1GbE vs 2.5GbE at 85% efficiency | type=bar | data=Recomputable, formula: TB × 8 × 1024³ ÷ (GbE × 10⁹ × 3600 × 0.85) | series=1GbE:14.9;2.5GbE:5.9]
The 2.5GbE port removes 9 hours from the same job without touching a single drive.
A single 2.5″ SATA drive sustains well above 1GbE in sequential writes. Two drives in RAID 1 do not change that ceiling.
So when the window is 15 hours, the drives are idle most of the time. They are waiting on the wire.
This is the part most buyers get backwards. They shop for faster drives first.
Buy the faster port first. Drives only set the pace once the link stops being the wall.
Four 8 TB drives give 32 TB raw. RAID 5 uses N−1, so 24 TB usable. RAID 6 uses N−2, so 16 TB.
That is 25% and 50% overhead respectively, and it applies before you count filesystem reserve.
[CHART: Usable capacity by RAID level, four 8 TB drives | type=stacked-bar | data=Recomputable, RAID5 = (N−1) × 8 TB, RAID6 = (N−2) × 8 TB | series=RAID5 usable:24;RAID5 overhead:8;RAID6 usable:16;RAID6 overhead:16]
Choosing RAID 6 over RAID 5 costs you 8 TB of the same four drives.
Surveillance is the workload that surprises people, because retention multiplies quietly.
Video storage in TB = cameras × bitrate (Mbps) × 86400 × retention days ÷ 8 ÷ 1024³.
Forty cameras at 2 Mbps for 30 days: 40 × 2 × 86400 × 30 = 207,360,000 Mb, and divide by 8 for bytes, then by 1024³. You land on 25.9 TB.
Put that on a 4-bay box with RAID 5 and you need four 12 TB drives to hold it with headroom, and on RAID 6 you are looking at four 16 TB.
[CHART: Required capacity by retention period, 40 cameras at 2 Mbps | type=line | data=Recomputable, TB = cameras × Mbps × 86400 × days ÷ 8 ÷ 1024³ | series=7 days:6.0;14 days:12.1;30 days:25.9;60 days:51.8]
Doubling retention doubles the drive bill, which is why 30 days is the usual compromise.
| Job shape | Data volume (TB) | Window @ 1GbE (h) | Window @ 2.5GbE (h) | Min bays @ RAID 5 |
|---|---|---|---|---|
| Daily incremental, 20-person office | 6 | 14.9 | 5.9 | 2 |
| Weekly full, same office | 18 | 44.6 | 17.8 | 4 |
| 40 cameras, 30-day retention | 25.9 | 64.1 | 25.6 | 4 |
| 40 cameras, 60-day retention | 51.8 | 128.2 | 51.3 | 4 |
All windows assume 85% link efficiency and a single concurrent job. Bays are the minimum to hold the volume with RAID 5 overhead applied.
The link speed only pays off if the box behind it can feed the wire. A four-bay desktop unit with a 2.5GbE port is the lowest-cost way to reach that ceiling — see the X4 4-bay NAS server specs for bay count and capacity limits.
Smaller sites can reach the same window with a two-bay chassis. A two-bay unit with a 2.5GbE port and a 60 TB ceiling handles most single-office backup sets.
If your existing unit already has 1GbE only, the decision is simpler: staying on 1GbE is the budget path, while a 2.5GbE chassis is the performance path. For buyers evaluating several suppliers at once, the OEM/ODM program page lists minimum order and customization scope.
For the underlying standard, the IEEE 802.3bz working group defines 2.5GBASE-T. Published drive workload ratings, which set the practical write ceiling, are documented in Seagate’s technical insight library.
If your window must fit inside 8 hours, this small office NAS backup window planning test gives you a clear order of operations.
First, check every hop. A 2.5GbE NAS port behind a 1GbE switch runs at 1GbE and that is the most common way buyers pay for speed they never receive.
Second, size the array to the retention, not to the drive count you happen to have.
Forty cameras at 30 days is 25.9 TB before RAID. Four 12 TB drives in RAID 5 is the honest minimum.
Third, decide RAID 5 or RAID 6 on rebuild risk, not on capacity.
RAID 6 costs you 8 TB on a four-drive array, and whether that is worth it depends on how long you can tolerate degraded operation.
Our own read: for a 20-person office running daily incrementals, the X4 with 2 × 2.5GbE is the right shape.
For a two-person studio, the A2 Pro at 1GbE is enough, because a 6 TB job finishing overnight is not a problem worth solving.
Match the port to the deadline. Do not buy capacity you will not fill.
Every window above scales linearly with efficiency, which makes it the most dangerous single input.
At 1.00 efficiency a 6 TB job on 1GbE would take 12.7 hours. At 0.70 it takes 18.1 hours.
That is a spread of 5.4 hours on the same hardware and the same data. Plan with the pessimistic number.
[CHART: Backup window sensitivity to link efficiency, 6 TB on 1GbE | type=line | data=Recomputable, hours = 6 × 8 × 1024³ ÷ (1 × 10⁹ × 3600 × eff) | series=0.70:18.1;0.75:16.9;0.80:15.9;0.85:14.9;0.90:14.1;1.00:12.7]
A five-point drop in efficiency costs you two hours on a single overnight run.
Small-file mixes are the usual reason efficiency falls. A backup of one 6 TB virtual disk streams at near line rate, while a backup of two million 4 KB files does not.
If your source is a file server rather than a hypervisor, test with the real mix before you size the link.
RAID level does not change the window at all, because parity is computed on the array side and the link never sees it.
[CHART: Backup window unchanged by RAID level, 6 TB set | type=bar | data=Recomputable, parity cost is internal to the array | series=RAID5@1GbE:14.9;RAID6@1GbE:14.9;RAID5@2.5GbE:5.9;RAID6@2.5GbE:5.9]
Pay for parity or pay for a faster link. They are separate budgets.
So the two decisions can be made independently. Choose RAID level on rebuild risk and capacity, then choose link speed on the window.
Every figure here assumes a single sequential job. Real offices run concurrent jobs, and concurrent jobs share the link.
We did not measure drive-level throughput, so the claim that disks are not the constraint rests on published SATA behaviour rather than our own bench.
The 0.85 efficiency factor is a planning assumption. Your protocol mix will move it.
The camera bitrate of 2 Mbps is a common mid-range figure, not a measured value from a specific camera model.
Use hours = TB × 8 × 1024³ ÷ (GbE × 10⁹ × 3600 × efficiency).
A 6 TB set on 1GbE at 0.85 efficiency gives 14.9 hours. Change the link speed and the whole result moves.
It depends on your deadline. If a 6 TB job must finish before staff arrive, 2.5GbE turns 14.9 hours into 5.9. If the job already fits overnight, the upgrade buys you nothing.
RAID 5 uses (N−1) × drive capacity.
Four 8 TB drives give 24 TB usable, with 8 TB consumed by parity. RAID 6 on the same drives gives 16 TB.
At 2 Mbps per camera, 40 cameras for 30 days need 25.9 TB before RAID overhead. On a four-bay RAID 5 array, plan for four 12 TB drives to keep headroom.
Plan on 0.85 for SMB over a clean gigabit path.
Overhead, protocol chatter, and small-file mixes all pull it down. Treat anything above 0.9 as optimistic for planning.
Tell us your application, capacity target and deployment scenario — our engineers return a configuration proposal with tiered pricing within one working day.
Key takeaways An FCC-recognized lab must hold the right scope for your device's rule part, not just any accreditation. A 2-bay NAS…
November 3, 2026Shared storage for deskside AI clusters refers to a storage appliance on the office LAN that holds datasets, checkpoints and embeddings for…
November 3, 2026A cloud storage subscription cost is the recurring fee a business pays to keep its files on a provider's servers, billed per…
Ask our export team for the OEM/ODM cooperation guide and a written quotation.
