networking, fiber optics, communication, technology, network, cable, connection, optic, speed, fiber, line, wire, fiber optics, fiber optics, fiber optics, fiber optics, network, cable, cable, cable, cable, cable, fiber, fiber, fiber. Fiber internet: 3 monitoring checks and 6 buying criteria
Photo by PawinG on Pixabay

Industry

Part of How fiber internet is delivered and what it changes

Fiber internet: 3 monitoring checks and 6 buying criteria

Fiber internet software means the portal, router firmware, and monitoring around your circuit. Three checks and six criteria to settle before you sign.

There is no such thing as fiber internet software in the strict sense. Glass carries light, and equipment at each end turns it into data, but the medium has no interface you log into.

What people mean by the phrase is the software around the circuit: the provider's portal, router firmware and configuration, and monitoring that tells you whether it does what you pay for.

Those three are worth choosing deliberately, because they decide what the circuit feels like far more than the glass does. What follows is what to look for in each and what to demand before you sign, whichever provider, router or monitoring service you end up with.

What to take away

  • Your router must forward at the circuit's full rate with the features you actually use switched on. Ask for that figure specifically.
  • Three monitoring checks, running all the time, are worth more than any speed test you ever run by hand.

The provider's portal

Every provider gives you some way to see your service. The useful ones let

you:

  • See the current state of the circuit and the equipment at your end, as the provider sees it, so you can tell whether a fault is visible from their side before you ring.
  • Raise a fault and watch its status, with the classification and clock visible.

What the portal must show

  • Circuit and equipment state, as provider sees it
  • Fault status with classification and clock
  • Usage over time, if it matters
  • Changeable settingsaddresses, contacts, notifications
  • Planned maintenance notices you will read

Ask to see the portal before you sign. A provider whose portal is a billing page with a phone number on it is telling you where faults will be handled.

Then put the fault handling into the contract itself: the service level the provider commits to, the point at which the repair clock starts, the credit you receive when the commitment is missed, and the minimum term you are signing up to. The wider comparison of providers is in provider comparisons; the portal is one of the things that comparison should include.

The router is where the circuit becomes real

A fiber circuit terminates in a small box the provider supplies. Behind that

sits your router, and your router decides everything: how fast the office actually goes, whether calls are prioritized, whether anything from outside can reach in, and whether a fault can be diagnosed.

Questions to settle about it:

Can it keep up?
A router has a rate at which it can move packets, and that rate falls sharply when inspection, filtering, or a VPN is switched on. Many routers cannot forward at the full rate of a modern circuit once those features are enabled. A box sold as a 1 Gbps router may publish only 200 Mbps to 400 Mbps with intrusion prevention and a VPN tunnel running, a typical spread in the small-office class. Ask the supplier for the throughput figure with the features you will use turned on, not the figure on the box. If they cannot give it, assume the small one.
Who manages it?
The provider, a contractor, or you. Each is fine. What is not fine is nobody knowing, which is the usual state. Managed by the provider means faster fault handling and less control; managed by you means the reverse. Write down which it is.
Where is the configuration?
Exported and stored somewhere the business controls, after every change. A router that dies takes its configuration with it, and rebuilding forwarding rules, prioritization and address assignments from memory is a bad afternoon.
When does firmware update?
Automatically at a time you chose, or manually in a window you control. Never at random. An update that drops the connection at eleven on a Tuesday morning is a self-inflicted fault.
Does it have what the circuit needs?
A port that matches the speed of the provider's box, prioritization for voice, and the ability to log what it is doing so that the monitoring below has something to look at.

Three checks that should run all the time

A speed test is a photograph. Monitoring is a film, and only the film shows

Three checks that run always

  • Outside-in checkexternal probe every minute
  • Throughput over timewired, fixed intervals, plotted
  • Delay and losssmall probes to a fixed point

the fault that happens at four in the afternoon on Thursdays.

  1. An outside-in check.Something outside the building that tries to reach your connection every minute and tells you when it cannot. This is the only way to know about an outage that happened while nobody was in, and it produces the timestamp the provider will want.
  2. Throughput over time.A regular measurement from a wired machine, at fixed intervals, plotted. Not to catch a slow moment but to see the shape of a week: the busy hour, the backup window, the day something changed. The Federal Communications Commission's own Measuring Broadband America program exists for the same reason: a single test at a convenient moment is not a measurement, and the shape over time is what an argument rests on.
  3. Delay and loss to a fixed point.A steady stream of small probes to somewhere stable, recording the round trip and any losses. This is the check that explains bad calls, because calls fail on variation and loss long before they fail on capacity. What those properties mean is set out in fiber internet, which also covers what the medium does and does not change about them.

All three exist as free tools, as features of routers, and as paid services. Among free, open-source options, SmokePing records delay and loss to fixed targets over time, and iperf3 measures throughput between two machines you control. Which you choose matters less than that they run continuously and keep their history. A workable setup is one probe a minute, a throughput test every 15 minutes, and history kept for at least 12 months, so this August can sit beside last August.

The device at your end is where most firewall and access-control guidance lands: if you enable filtering or a VPN, document what you allow in and out so a fault report can say what changed.

Change discipline

Most faults follow a change, and most changes on a small network are

unrecorded. Three habits:

  • A change log for the router and the provider's equipment: what changed, when, who, why. One line each.

Change discipline habits

  • Change logwhat, when, who, why
  • Configuration export after every change
  • Maintenance window for firmware and drops

Six criteria for any router, tool or managed service

Whatever you are choosing, a router, a monitoring tool, or a managed

service, hold it against this list:

  • A forwarding rate quoted with inspection, filtering and VPN switched on.
  • Voice prioritization that holds up in the busy hour.
  • A configuration you can export and restore yourself.
  • Firmware updates that run in a window you set.
  • Logs that leave the box and land somewhere you can read.
  • Fault visibility from the provider's side, seen through the portal.

Choosing without naming names

  • Forwards at circuit rate with your features on
  • Exports configuration you can store and restore
  • Updates when you say, not when it likes
  • Logs enough to answer what happened at 15:42
  • Manageable by someone other than the installer
  • No subscription needed for the basic job

A product that fails two of these will cost more in time than it saved in money, however it is priced. The setup records that make any of this usable in a fault are described in setup and troubleshooting.

Common questions

Can I use the router the provider gives me?

Often, for a simple office. Check whether it can prioritize voice, whether you can export its configuration, and whether you or they control its updates. If the answer to all three is no, put your own router behind it and have theirs do as little as possible.

How much monitoring is too much?

Any monitoring nobody looks at. Three checks that produce a graph someone glances at weekly beat twenty checks that send alerts everyone has muted.

Does the circuit need anything from me to stay fast?

Nothing. It needs your router not to be the limit, your wiring inside the building to match, and your busy-hour demand to fit within what you bought. The circuit itself is the one part that does not degrade with use; the contract side of that is covered in business internet.

Should monitoring be on the provider's box or mine?

Yours, or somewhere independent. Monitoring that lives on the provider's side stops reporting at exactly the moment you need it to.

More in Industry

Features

Fiber internet security at the edge

Fiber internet security starts at the edge: learn what fiber does not protect and how to manage exposed services, network separation and access.

Latest from Costs Desk