Enter how much data you’re moving, the connection, and the drives at each end. The calculator gives you the realistic transfer time, not the marketing one. Here’s the logic behind it.
Estimated transfer speed is the slower of usable network throughput and the disk. The calculator first applies the selected network efficiency to the link rate, then compares that usable rate with the drive’s sustained speed. A 10GbE link at 92% efficiency supplies an estimated 1,150 MB/s; a 180 MB/s hard drive still limits that transfer to 180 MB/s. The network overhead setting does not reduce the drive preset a second time.
The bottleneck readout is the useful part. It tells you whether spending money on faster networking would actually help. If the drive is the limit, a 10GbE card changes nothing until you add disks or switch to SSDs. If the link is the limit, faster networking pays off directly.
Time is just size divided by speed. Once the effective speed is set, the math is exact: a 1 TB file at 280 MB/s takes about an hour. The per-link table shows the same transfer across every common Ethernet speed so you can see where the gains flatten out for your storage.
Sizes use decimal units (1 TB = 1,000 GB) to match how both drives and network speeds are quoted.
Worked examples
These are this calculator’s own outputs, computed with the same formula the tool runs in your browser — so you can see a real answer without touching a single input.
The page’s own default state: backing up an 8 TB media library off a single 3.5" hard drive to the NAS over a 2.5GbE link, at typical efficiency. This is the exact result a reader sees the instant the page loads
- Inputs: Amount = 8, Unit = TB, Source/target drive = Single 3.5" HDD (180 MB/s), Link speed = 2.5GbE (312.5 MB/s), Efficiency = Typical (0.92)
- Result: Headline: 12 h 20 m. Estimated transfer speed 180 MB/s, Bottleneck “Drive”, theoretical line time 7 h 6 m. The 2.5GbE link supplies an estimated 288 MB/s after network overhead, so the 180 MB/s drive sets the pace. The table shows 2.5GbE, 5GbE, 10GbE and 25GbE all tied at 12 h 20 m.
Moving a 2 TB project archive between two NVMe Gen4 workstations over 10GbE at typical efficiency — the case where the network, not the storage, is the limit
- Inputs: Amount = 2, Unit = TB, Source/target drive = NVMe Gen4 SSD (7000 MB/s), Link speed = 10GbE (1250 MB/s), Efficiency = Typical (0.92)
- Result: Headline: 28 m 59 s. Estimated transfer speed 1.15 GB/s, Bottleneck “Link”, theoretical line time 26 m 40 s. Going from 10GbE to 25GbE would cut it to 11 m 35 s because the NVMe can still feed it.
The classic homelab question: seeding 4 TB onto a four-bay HDD NAS over a new 10GbE link, at typical efficiency — does the 10GbE card actually earn its money?
- Inputs: Amount = 4, Unit = TB, Source/target drive = HDD RAID array, 4-bay (450 MB/s), Link speed = 10GbE (1250 MB/s), Efficiency = Typical (0.92)
- Result: Headline: 2 h 28 m. Estimated transfer speed 450 MB/s, Bottleneck “Drive”, theoretical line time 53 m 20 s. The table flatlines at 2 h 28 m from 5GbE upward, so 10GbE and 25GbE buy nothing over 5GbE for this array.
Assumptions and sources
- Ethernet link rates are the interfaces’ own data rates converted to decimal bytes (10GbE = 10 Gbps = 1,250 MB/s), because 64b/66b line coding sits below the advertised Ethernet rate. USB and Thunderbolt advertise a signalling rate instead, which has to be de-rated for line coding before it is payload: USB 3.2 Gen1’s 5 Gb/s is 8b/10b, so 500 MB/s rather than 625; Gen2 and Gen2x2 are 128b/132b, so 1,212 and 2,424 MB/s rather than 1,250 and 2,500; and Thunderbolt 3/4’s 40 Gb/s is a shared bidirectional link budget, so direct-attached storage rides tunnelled PCIe 3.0 x4 at 3,938 MB/s. The decimal-versus-binary labeling gap the FAQ describes is a separate thing - the drive-industry convention documented in Seagate’s capacity explainer.
- The network efficiency presets (typical = 0.92, reflecting the 8-12% TCP-plus-SMB/NFS overhead the FAQ describes) are TechFuelHQ working estimates from real transfers, stated so you can pick a different preset when your stack does better or worse. The formula is
min(drive speed, link speed × network efficiency); the drive presets already describe sustained throughput and are not discounted by network overhead. - The drive-speed presets (180 MB/s single HDD, 450 MB/s 4-bay array, 550 MB/s SATA SSD, 7,000 MB/s Gen4 NVMe) are TechFuelHQ typical sustained figures; your drive’s spec sheet, fill level, and file mix all move them.
- The small-file penalty is deliberately not modeled, as the FAQ states - it is workload-specific and can outweigh every other factor on folders with thousands of tiny files.