Algeria AgencyNetwork & connectivity
The line is down: establishing which side, and what comes next
Five minutes decide whether the fault is yours or the operator’s. What to record before calling, how to escalate, and what a second line really covers.
The companion article describes the chain: seven segments, the weakest link, the one-page diagram, addressing, and what you can demand of an operator at the moment of signing.
This one begins when the chain stops. The question is no longer how to design it, it is: which side of the wall is the fault on, and what do you do in the twenty minutes that follow.
One point about scope, because the neighbouring switch article handles the same subject from the other side: here we are talking about the line you rent, outside the building. What is inside — the loop, the second address server, the small switch under a desk — is handled there.
The boundary is not a writing convention. It is the first thing to establish on the ground, it takes five minutes, and until it is established everybody is discussing a fault nobody knows the owner of.
It all starts as a question of boundary
An outage always produces the same scene. Somebody calls the operator, somebody else restarts the box, a third person unplugs cables, and nobody has established whether the fault is yours or theirs.
That is where most of the time goes, and it is not a matter of competence: the two faults have exactly the same symptom seen from a workstation. "The internet is not working" describes some thirty different situations spread across both sides of the wall.
The cost of not settling it is twofold. You call an operator about a fault that is yours, and wait forty minutes to be told the line is fine. Or you search your own premises for two hours for a fault on the network, which would have been reported in six minutes.
The good news is that settling it is easy, free, and takes five minutes and no tools. Section 2 gives the sequence, and it is the only thing in this article worth knowing by heart.
One organisational consequence before going on: decide now who does those five minutes. Nine o’clock in a shop during an outage is not the moment to appoint somebody, and the natural reflex — everybody deals with it — is the most reliable way to establish nothing at all.
Three different lines, three different failures
Before searching, you need to know what you have. Three technologies share almost the whole Algerian fixed park, and they do not fail in the same way.
Copper is the oldest and the most widespread, and it goes down mainly through damp and roadworks: the failure is clean, it affects the neighbourhood, and it is repaired by a physical visit.
Fibre cuts very rarely and cuts totally when it does: a severed cable does not degrade, it stops. The good news is that the failure is unambiguous; the bad news is that there is nothing to do from your side.
The third deserves attention because it is large and poorly understood: close to three fixed subscriptions in ten actually run over a radio link. A radio link almost never cuts — it degrades, at busy hours, in rain, or when a building goes up in the line of sight.
The practical consequence is this: if your line is a fixed 4G one, the failure you need to learn to describe is not an outage but a slowness, and everything you record about a clean cut will be useless. It is also why section 6 asks you to record quality rather than only state.
ARPCE, internet market observatory, second quarter of 2025
Establishing the side in five minutes
The sequence is four moves and they go in this order, with none skipped. Each eliminates half the problem.
First move: look at the operator box’s indicator lights before touching anything. The next section says which ones matter. If the synchronisation light is off or blinking, the fault is outside and you have finished in thirty seconds.
Second move: plug a laptop directly into the box, with a cable, having unplugged everything else. If that works, the line is fine and the fault is yours — you move to the neighbouring article on switches.
Third move: test using a colleague’s phone as a mobile hotspot. That repairs nothing and answers a different, useful question: is the problem the line, or is the service you are trying to reach itself down?
Fourth move, and this is the one nobody makes: ask the shop next door. A neighbourhood outage is confirmed in a minute through a window, and that information completely changes what you will say on the phone — and how long the operator takes to believe you.
The light that does not mean what people think
Boxes have three or four lights and almost nobody knows which to look at. That is not a detail: it is what separates a six-minute call from a forty-minute one.
The first light that matters is synchronisation, often named after the line’s technology. It says one thing: is your box talking to the operator’s equipment at the far end. Off or blinking, the fault is outside, full stop.
The second is the internet or service light. It says something else: the line is established, and has the operator authorised you out. Synchronisation on and service off is a very particular and very common case — the line is physically fine and there is a subscription, connection-password or operator-side configuration problem.
That distinction is the most useful thing in this article, because it names a fault the first level of support resolves in two minutes if they know what you are talking about, and with a technician’s visit if they do not.
The third light, the wireless one, has nothing to do with the line and is nevertheless the one everybody looks at. It shows only that the box is transmitting Wi-Fi, which it keeps doing perfectly during a total outage — which explains the "but the Wi-Fi is working" heard at every failure.
What to have to hand before dialling
A prepared call takes six minutes and an improvised one takes forty. The difference is not in what you know how to do, it is in six pieces of information written on paper before you pick up.
The line number or subscription identifier, which is not your phone number and is on the invoice. The exact time the fault started. The state of the lights, named one by one.
Then: total outage or degradation. What you have already tested — the computer plugged in directly, especially, because that is what proves your own installation is not at fault. And whether the neighbourhood is affected.
Write those six things on a laminated card stuck to the box, today, with the support number. The person who calls will almost never be the one who read this article, and a card on the box is the only place they will look.
One last point, true of every operator: at the end of the call, ask for the fault reference and write it down. Without it, your outage does not exist in their system, and the second call starts again from zero with somebody else.
The first level, and how to get past it politely
The person at the other end follows a script, and that script exists because the majority of the calls they take are internal faults at the customer’s end. They cannot know otherwise until they have asked their questions.
The right way to shorten it is not to refuse the script, it is to get ahead of it. Give in your first sentence the state of the lights, the fact that you plugged a computer in directly, and the start time. You have just answered three quarters of what they had to ask.
What to accept redoing even if you have already done it: restarting the box while they are on the line. They need to see it from their side, it takes two minutes, and arguing costs more time than doing it.
What not to accept: the call ending without a record being created. If the conclusion is "it should come back", ask for the fault to be registered anyway. An unregistered outage did not happen, and that is exactly what is missing from the file in section 6.
And a point of method that costs little and returns a lot: note the person’s name and the time of the call. That is not in order to complain about them, it is because the third call of the week becomes a different conversation when you can say what was done on the first two.
The file: what turns a complaint into a case
"Our line is bad" is an opinion. "Eleven outages in three months, seven of them between two and four in the afternoon, with the fault references" is a case, and the two sentences get very different answers.
The register fits on one sheet and has five columns: the date, the start time, the duration, the state of the lights, and the fault reference obtained. Nothing else, because a complicated sheet does not get filled in.
Distinguish it from the record the neighbouring switch article asks for. That one correlates an internal fault with what is happening in the premises; this one demonstrates a frequency to a third party. They live in the same place and serve different audiences.
Three months is the right span before doing anything with it. Below that you are describing a bad week; above it you are describing a situation, and that is the only level of evidence that moves anything without going through a personal relationship.
The register has a second use, less obvious and more useful day to day: it tells you whether the line is degrading or your memory is. We have several times seen a line described as "catastrophic" produce three outages in a quarter, with the real problem indoors.
Escalation, when the first level has nothing left to offer
The neighbouring article covers what you can demand at signature. What follows is the other half: what you do when what was promised does not arrive and the support number goes round in circles.
The first step is writing, and most businesses reach it far too late. A written complaint, lodged at the commercial agency you belong to, with the fault references from the register, opens a matter that customer service cannot close with a word.
Keep a stamped copy, or send it with proof of receipt. That is not distrust: it is what will let you say, two months later, on what date the complaint was lodged, without that depending on anybody’s goodwill.
The second step is the sector regulator. The regulatory authority handles subscriber complaints that have had no answer from their operator, and the existence of that route often changes how the first step is handled on its own.
Two honest limits, because this article would be no use to you without them. None of this speeds up a physical cable repair, which will take as long as it takes. And escalation is a matter of weeks, which means it never replaces the next section — it deals with a line that is persistently bad, not with Tuesday.
The second line: what it covers and what it does not
A second line is the only real answer to an outage, and it is bought the wrong way round in almost every case we see.
The commonest mistake is taking a second subscription from the same operator, on the same technology, arriving by the same route. That doubles the bill and protects against nothing: both lines go down together, because what goes down is almost always the shared segment.
So the rule is that the second line must differ by technology and if possible by operator. The combination that works best here is a main fixed line and a 4G box as backup, precisely because the two do not fail for the same reasons.
What to know before buying, and nobody says it: a 4G backup line has a monthly volume, and that volume runs out in two days when the whole business is working on it. Size it for what has to continue, not for normal operation.
And one thing the second line does not cover at all: the public address changes when you switch. Everything that depends on your address — remote access, a tunnel to another site, a payment terminal declared on a fixed address — stops working at the precise moment you need it. That is the subject of the next section.
Switching over, and the fact that nobody has ever tried it
A backup line that has never been tried is a subscription, not a protection. It is the same rule as a backup never restored, and it fails the same way: everybody is certain it will work.
The first choice is manual or automatic, and it is less obvious than it looks. Automatic switches in seconds with nobody intervening; manual requires knowing where the cable is and who moves it, which is a problem when it happens at six on a Friday evening.
The second choice matters more and is not set in a box: what happens on the way back? A switchover that does not return by itself leaves a business on its backup subscription for three weeks, until the volume runs out abruptly in the middle of a day.
The rehearsal takes twenty minutes and is done on a day when everything is fine: unplug the main line, watch what switches, open each of the three or four things that have to continue, then plug it back in and watch what returns. Write down what broke.
What this test almost always reveals is the same thing: the till works, the mail works, and the payment terminal or the remote access does not. Discovering that on a quiet Tuesday costs twenty minutes; discovering it during an outage costs the day.
What has to keep working without the line
The question is not making everything work without the internet, which is impossible and expensive. It is deciding in advance what has to continue, and that decision is commercial rather than technical.
Make the list calmly, in a twenty-minute meeting, and hold it to four lines. Beyond four it is no longer a list of priorities, it is an inventory, and nobody will use it during an outage.
For most of the shops and workshops we equip, the four are the same: take payment, take calls, reach the day’s orders, and tell customers. Everything else waits, and saying so in advance avoids deciding it in a panic.
Each of those four lines is then handled separately, and the handling is not always technical. Taking payment may mean a notebook; telling customers may mean a mobile phone and a printed list. A solution on paper is not a defeat, it is the only recourse that depends on no line at all.
The list goes up next to the box’s card, and it carries a date. Its real function is not to organise the outage: it is to stop the outage being organised by whoever is most worried.
What we do, and what we refuse to do
What we refuse first: selling a second line from the same operator on the same technology. It sells very well, it doubles a bill every month, and it protects against nothing since both lines share the segment that fails.
We also refuse to promise a restoration time. We hold none of the links that decide that time, and a provider announcing four hours on a line they do not own is selling you somebody else’s peace of mind.
And we refuse, most frequently of all, to invoice a call-out for an operator fault. The five minutes in section 2 are for that too, and we do them on the phone with you rather than on site.
What we do fits in half a day: the card on the box, the register started, the second line sized against the list of four priorities, and the switchover rehearsed once in front of you, the way back included.
And one thing to do today without us, in ten minutes: find your line identifier on an invoice, write it with the support number on a label, and stick it on the box. The next outage will be handled by somebody who did not have to go looking for the invoice.
Frequently asked questions
How do we know whether the fault is ours or the operator’s?
Four moves and five minutes. Look at the box’s synchronisation light before touching anything — off or blinking, the fault is outside. Plug a computer directly into the box: if that works, the fault is yours. Test using a phone as a hotspot. And ask the shop next door, which confirms a neighbourhood outage in a minute.
The Wi-Fi works but there is no internet. Why?
Because the wireless light has nothing to do with the line: it shows only that the box is transmitting, which it keeps doing perfectly during a total outage. The two lights that matter are synchronisation, which says whether your box is talking to the operator, and service, which says whether the operator is letting you out.
What should we have ready before calling support?
Six things: the line identifier, which is on the invoice and is not your phone number; the exact start time; the state of the lights named one by one; total outage or degradation; what you have already tested, especially the computer plugged in directly; and whether the neighbourhood is affected. A prepared call takes six minutes, an improvised one forty.
The operator is doing nothing. What is the recourse?
Writing first, and much earlier than most businesses manage: a complaint lodged at the commercial agency, with the fault references from your register, and a stamped copy. Then the sector’s regulatory authority, which handles subscriber complaints left unanswered. Two honest limits: none of this speeds up a cable repair, and escalation is counted in weeks.
Do we need a second line, and which one?
Yes if a day of stoppage costs you more than twelve months of a backup subscription. And it has to differ by technology, ideally by operator: a second line of the same type arriving by the same route doubles the bill and fails at the same time as the first. The combination that works here is a main fixed line and a 4G box as backup.
Our backup has never been used. Is that a problem?
It is the commonest case and it is a subscription rather than a protection. Rehearse on a quiet day: unplug the main line, open each thing that has to continue, plug it back in, and watch what returns. What the test almost always reveals is that the till and the mail work and the payment terminal or remote access does not — because the public address changed.
Where we come in
A label stuck on the box saves twenty minutes on the day, and it goes on tonight. It does not say which side of the line went down.
- We show you the three checks that establish where the break sits.
- We size the backup on four written priorities, never on a bandwidth figure.
- We rehearse the switchover once, on a day when everything works.
When the fault is at the operator we can speed nothing up, and we will not take your money to wait beside you.
Read next
Networks: the cheapest layer, and the one that produces half the faults
A network is a chain: it is worth what its worst segment is worth. That is why it gets bought in the order the signal travels.Moving the business: what to order three months ahead, and the weeks you exist at two addresses
A move tests your whole system at once, on a date a landlord set. The long pole is never the van.Telephony: the number is the asset, not the system
Your customers know a number, not a piece of equipment. Everything decided here has to protect that number and what happens when somebody dials it.
Let us talk about your project
A free audit, no commitment: we look at your online presence and tell you what is holding it back.