Because a particle far too small to see will attenuate the signal, and contamination causes more optical faults than every other cause combined.

An optical core is a few microns across on single-mode fibre. A speck of dust, a fingerprint or a residue of oil sits directly in the light path and scatters or absorbs a meaningful proportion of it.

The result ranges from a connection that will not establish, through one that establishes with errors, to one that works and fails intermittently as temperature changes - the last being the hardest to diagnose.

Worse, mating a contaminated connector transfers the contamination to the other end face and can permanently damage both by pressing the particle into the glass.

The discipline is simple: inspect before mating, clean if anything is visible, re-inspect after cleaning, and cap everything that is not connected. A kit costs very little against the time spent chasing a single unexplained optical fault, and it should be standard equipment wherever fibre is handled.

An inspection scope, a click-type cleaner for connectors, cassette cleaners, MPO tooling if parallel optics are used, and lint-free wipes with the correct solvent.

The inspection scope is the essential item, because cleaning without inspecting is guesswork. A video scope showing the end face at magnification reveals contamination, scratches and pits, and confirms whether cleaning worked.

Click-type pen cleaners handle connectors in place, including inside adaptors and bulkheads - which is where most connectors need cleaning and where wipes cannot reach.

Cassette cleaners handle patch lead ends quickly.

MPO connectors need their own cleaners and a scope capable of showing all fibre positions - single-fibre tooling does not work.

Wipes and solvent handle stubborn contamination that dry cleaning does not remove; use only fibre-grade solvent and lint-free material, since ordinary alcohol leaves residue and paper leaves fibres.

Keep the kit where the fibre is, not in a cupboard in another building.

Proving that a port and its module work, by returning the transmitted signal to the receiver - which isolates a fault to the equipment or the cabling.

An optical loopback is a short fibre path in a connector housing that returns the module's transmission to its own receiver. If the port comes up with a loopback fitted, the switch port and the module are both working and the problem is in the fibre plant or at the far end.

A copper loopback does the same for electrical interfaces.

That single test removes most of the guesswork from an optical fault. Without it, diagnosis usually proceeds by swapping modules and patch leads, which is slower and introduces new variables.

Loopbacks must match the connector type and, for optical types, often include attenuation - because a module's transmitter fed straight into its own receiver at full power can exceed the receiver's maximum. Using an unattenuated loopback on a long-reach module gives a false failure or damages the receiver.

When the received power is too high - typically a long-reach module used over a short distance.

A receiver has a working range with a maximum as well as a minimum. A module designed for 40 or 80 kilometres, connected over 200 metres, delivers far more power than the far receiver can handle. The result is a connection that will not establish, or one that produces errors, and the cause is the opposite of what people expect.

A fixed attenuator of the appropriate value, fitted in the path, brings the level into range. Values are chosen from the difference between the measured receive power and the receiver's specification, with margin.

The module's own diagnostics report receive power, which is how the required value is determined - measure rather than guess.

Variable attenuators are used in testing to establish a link's margin.

The better answer where possible is to use a module matched to the distance. Attenuators are the remedy where a long-reach module is what is available or where the same part is standardised across an estate.

Yes, with a programming device - and it is how suppliers code modules and how spares are made flexible.

A module's identification is held in a small memory. Programming devices read and write it, so a module coded for one platform can be re-coded for another.

That has real practical value. A stock of modules can be held generically and coded as needed rather than stocking separately per platform, and a module recovered from decommissioned equipment can be reused elsewhere.

The optical characteristics are unaffected - coding changes only what the module says about itself, not what it does.

The cautions are that re-coding a module supplied by an equipment vendor may affect support arrangements, that the correct code strings must be used or the switch will still reject it, and that the module must genuinely be compatible - coding a module for a platform whose electrical requirements it does not meet produces a link that fails in confusing ways rather than one that works.

A limited set, and only where the electrical interface allows - QSFP to SFP adaptors being the common useful example.

An adaptor that lets an SFP or SFP+ module be used in a QSFP port exists on some platforms, using one of the quad port's four lanes. It is useful where a switch has only quad ports and a single low-speed connection is needed.

OSFP to QSFP adaptors allow existing QSFP modules to be used in OSFP ports, which is how that format provides backwards compatibility.

What does not exist is an adaptor between formats whose host electrical interfaces differ fundamentally - XFP to SFP+ being the clearest example, since the division of function between host and module is different.

Support is platform-specific: an adaptor that works on one switch may not be recognised on another. Check the vendor's compatibility list rather than assuming, and test before relying on it in a design.

Spare modules of each type in use, patch leads, dust caps, a cleaning kit, loopbacks and any attenuators the design depends on.

Modules should be held for each combination of format, reach and coding in service - a spare of the wrong reach is no help. Record what is installed so the spare list stays accurate.

Patch leads in the right connector types and fibre modes are needed more often than modules, because leads are handled, bent and damaged.

Dust caps are trivial and constantly lost, and their absence is what causes the contamination the cleaning kit then has to remove.

Cleaning tooling and loopbacks belong in the same kit, because they are what make a fault diagnosable at three in the morning.

Where attenuators are part of a link's design, hold spares and - more importantly - record on the documentation that the attenuator is there. A link rebuilt without it will fail, and nobody will understand why.