Why does a gigabit link never transfer a full gigabit?

A gigabit is a line rate in bits, a file is measured in bytes, and framing and protocol headers occupy part of every transmission. What arrives is then limited by the slowest element in the chain rather than by the link.

Three separate things are being confused, and they subtract in that order.

Bits and bytes are not the same unit

Network rates are quoted in bits per second. File sizes and copy speeds are quoted in bytes. There are eight bits in a byte, so the first step in comparing them is a division: one thousand megabits per second is one hundred and twenty-five megabytes per second before anything else has happened.

That single conversion accounts for most of the shock. A transfer meter that appears to be running at an eighth of the link rate is running at the link rate, in the other unit.

Framing and headers take their share

Every frame on the wire carries addressing and error checking around the data it holds, and there is a mandated gap between frames. Above that sit the protocol headers of whatever is being carried. None of that is waste — it is what makes the data arrive at the right place intact — but it does occupy part of the line rate, and the payload is what remains.

How much remains depends on the size of the frames in use, so it is not one fixed number. It is why the useful figure is always the one measured over a real transfer of a real file rather than a subtraction done on paper.

The slowest element decides the rest

The link rate is a ceiling. What a transfer actually achieves is set by whatever in the chain is narrowest, and on a home network the link is rarely it.

Storage at either end. A spinning disk has a sustained rate of its own, and a solid-state drive publishes a burst figure that applies while a fast write cache absorbs data and a lower sustained figure once it is full. A long copy runs at the second one.

The protocol and the machine. File sharing, encryption, compression and anti-virus scanning are all work done by a processor, and a small machine can be the limit rather than the wire.

Many small files. A folder of thousands of small files spends most of its time on per-file work rather than on moving bytes, and it will never approach the rate a single large file reaches over the same link.

A USB enclosure in the path. The bridge inside it has a rate of its own, and the drive behind it cannot exceed it.

What is not the reason

The cable, usually. A copper channel that meets its grade carries the rate it is specified for, and a link that has come up at gigabit has already negotiated it. Cabling shows up as a link that comes up at a lower rate or that reports errors, not as a slow but stable copy.

How to find your own restriction

Copy a single large file rather than a folder, both ways, and watch where it settles. Then copy the same file from local storage to local storage on each machine. If the local copy is no faster than the network copy, the storage is the limit and the network is innocent — which is the most common finding, and the one that saves the price of a switch.

The manufacturer’s own documentation comes first, and this page is not a substitute for it. Every rate, capacity, class figure, endurance rating, port speed and power figure on this page is reported from the manufacturer's own documentation, with its unit written as published, and named as such — none of it is measured here. A stated capacity is reported as the maker counts it, and is never restated in another unit, because the two are not the same figure. And the statement repeated rather than assumed: a class figure such as AX3000 or BE9300 is the arithmetic sum of the theoretical peak rates of separate radios, a quantity the standard defines rather than one a link produces. No single connection reaches it, and adding the bands together is exactly what makes it large. What a given link carries is settled by the slowest part of the chain, which no comparison page can know.