Skip to main content
Mini PC Lab logo
Mini PC LabMini PCs for Homelabs

Network Transfer Time Calculator

How long will that copy take, and would a faster network actually help? This works out both ends of the transfer and tells you which one is the limit, because upgrading the wrong one is the most common waste of money in a homelab.

Time to move 2000 GB

2 hours

at an effective 281.3 MB/s

Bottleneck: the network

2.5 Gigabit Ethernet after overhead
281.3 MB/s
Hard drive array x4
800 MB/s

A faster network link would speed this up.

  • The network is the limit here. The storage can supply 800 MB/s but the link only carries 281.3 MB/s, so a faster link would help.
  • Sequential throughput roughly scales with the number of data drives. Random workloads do not scale this way.
  • This assumes one large sequential transfer. Many small files are dramatically slower on any medium, and worst on mechanical drives.

Your 10GbE is probably not the problem

A transfer runs at the speed of its slowest component, and on a homelab NAS that is usually the disks. A single modern hard drive sustains somewhere around 200 megabytes per second on sequential reads. A 2.5 gigabit link carries about 280 after protocol overhead. So one drive cannot even fill 2.5GbE, and the 10 gigabit card you were considering would sit idle 90% of the time.

This flips once the array gets wide. Sequential throughput roughly scales with the number of data drives, so a six-drive array can genuinely saturate 10GbE while a two-drive mirror cannot. That is the honest case for 10GbE in a homelab: enough spindles, or flash.

The calculator names which side is the limit and says outright whether a faster link would change anything. When it says no, believe it and spend the money on drives.

Where the numbers come from

Link speeds are standards. A gigabit link carries 1000 megabits per second, which is 125 megabytes per second before any overhead, and large sequential transfers over wired Ethernet rarely exceed about 90% of line rate once protocol overhead is counted. That efficiency is adjustable above, because it varies with protocol and tuning.

Storage figures split into two kinds and the tool treats them differently. SATA and PCIe rates are interface ceilings: SATA III signals at 6 gigabits per second, so no SATA SSD exceeds roughly 550 megabytes per second regardless of what the box claims. Mechanical throughput is a typical value, not a ceiling, because it varies by model and even by position on the platter. That is why it is an editable field rather than a number we assert. Check your drive's specification sheet and put the real figure in.

Sequential is the best case

Everything here assumes one large file moving in sequence. Thousands of small files behave completely differently, because each one carries filesystem and protocol overhead, and on mechanical drives the heads have to seek between them. A backup of a photo library can run an order of magnitude slower than this calculator suggests, and no network upgrade fixes that either.

The other thing that will not show up here is a slow intermediary. A cheap switch, a negotiated-down link, or a USB enclosure between the drive and the machine becomes the bottleneck without announcing itself.

Related

To size the array itself, use theNAS capacity calculator, and for the memory the machine needs the TrueNAS hardware calculator. On the hardware side see our10GbE mini PC picks, the2.5GbE versus gigabit comparison, and the NAS mini PC roundup.