A great deal, and predictably - hundreds of megabits at a couple of hundred metres falling to tens of megabits at a kilometre or more.

The technologies used are the same family as broadband over telephone lines, and they behave the same way: the achievable rate falls steadily as the line gets longer because the signal attenuates and the noise does not.

Manufacturers publish rate-against-distance curves for their products, and those curves are the specification. A figure quoted without a distance is meaningless.

Cable quality matters as much as length. Structured cabling performs better than old telephone cable; a single clean pair performs better than one in a crowded multi-pair with crosstalk; joints, moisture and damage all reduce the rate.

So estimate conservatively and test on the actual cable before committing. The same product on two apparently similar 600 metre runs can give quite different results, and the difference is the cable rather than the equipment.

Almost any twisted pair, and some products work over coaxial or two-wire alarm cable - which is often the point.

Most extenders need only a single twisted pair, so a spare pair in an existing telephone multi-pair, an old alarm circuit, or one pair of a structured cable is sufficient. That is what makes them useful in buildings where nothing new can be installed.

Some products work over coaxial cable, which suits sites with legacy camera or television distribution already in place.

What matters is the electrical condition rather than the category: continuity, insulation, absence of water, and freedom from noise sources. An old pair running alongside a power cable in a shared duct may perform poorly regardless of length.

Test the actual pair before designing around it, and identify a spare in case the first performs badly - the usual outcome of a disappointing installation is that a different pair in the same cable works considerably better.

Many can, and that capability is frequently the reason for choosing one product over another.

A camera or access point at the far end of a long run needs power as well as data, and if there is no mains supply at that position the extender must provide it.

Extenders with PoE pass-through take PoE in at the near end - from a PoE switch or an injector - carry power over the extended link, and present a PoE port at the remote unit.

The limits are real: the power available at the far end is less than what went in, and the standard supported may be lower than the source's. Check the remote unit's output class against the device's requirement, not just the input class.

Some units instead need local power at the remote end, which removes much of the benefit.

Where a device with a heater is involved - most outdoor cameras - confirm the figures at the cold-weather peak draw rather than the typical, since that is when the link matters most.

That is set by the technology, and it matters more than people expect for cameras and uploads.

VDSL-based extenders are frequently asymmetric, with a higher downstream rate than upstream, reflecting their broadband origins. If the traffic is predominantly in one direction that can be arranged to suit - but the direction must be checked against the application.

A camera at the remote end sends video toward the centre, which is upstream from the remote unit's point of view. An asymmetric extender configured the wrong way round gives that traffic the slower path, and the result is poor video quality with plenty of apparent bandwidth in the unused direction.

SHDSL-based products are symmetric by design and are usually the better choice for camera and general network use.

Many products allow the profile and direction to be configured, so the fix is often a setting rather than different hardware - but only if somebody checks.

Fibre is better in every respect except that it has to be installed - which is exactly the constraint extenders address.

Fibre gives full speed regardless of distance up to kilometres, no electrical path between buildings, immunity to interference, and capacity for future speed increases. Where fibre can be installed it is always the correct answer.

The extender's case is entirely about the existing cable. In a listed building, a site where no new containment can be run, a lease that prohibits works, or a temporary requirement, the choice is between an extender and nothing.

It is also a legitimate interim measure: an extender restores service now while fibre is planned and installed.

What should be avoided is choosing an extender to save the cost of a fibre installation on a permanent link. The performance ceiling is fixed and low, and the site will be living with it for as long as the building stands - by which time the fibre would have paid for itself several times over.

In practice yes. The line technology is standardised but the implementations do not interoperate reliably.

Although VDSL and SHDSL are standards, extender products use vendor-specific profiles, rate adaptation and management, and a pair from two manufacturers will frequently either fail to link or link at a poor rate.

Products are also usually asymmetric in role: one unit is the master or central unit and the other the remote, and two of the same role will not link at all. Ordering two identical units without checking is a common mistake.

Buy them as a matched pair, and keep a spare pair rather than a spare unit where the link matters - a failed master cannot be replaced by a spare remote.

Multi-port central units exist for sites with several extended links, aggregating them into one managed device, which is a much tidier arrangement than several standalone pairs.

Read the line statistics on the units themselves - rate, signal to noise margin and error counts tell you almost everything.

Most extenders report the negotiated rate and a signal-to-noise margin. A healthy link has a comfortable margin; one with a small margin will be intermittent as conditions change, and the rate will fluctuate.

A falling rate over time indicates the line deteriorating - water ingress into a joint is the classic cause, and it is often seasonal.

High error counts with an adequate rate usually mean an interference source near the cable, which may be new equipment installed nearby rather than anything about the link.

If the statistics are poor, try a different pair in the same cable before suspecting the equipment - the improvement is frequently substantial.

And confirm both units are set to the same profile. A mismatched configuration produces a link that establishes at a low rate with no obvious error, which is easily mistaken for a distance limitation.