Why two lines on the same carrier behave differently
Why two lines on the same carrier behave differently
It usually arrives as a swap request. Line three is worse than line one, can I have a different one.
Both lines are on the same carrier, activated the same week out of the same batch of cards, sitting on one shelf in a room in my house close enough that I can put a hand on both at once. Same SIM cost, about ten dollars a month each whether anyone uses them or not, same modem depreciating at roughly a dollar fifty a month.
The customer is right that they differ. Worth establishing first, because the reflex on my side of the desk is to open a dashboard, see green across the row, and reply that the line is fine. I have done exactly that and been wrong.
How a platform decides whether an address looks trustworthy is a subject of its own, and so is how the three Singapore networks compare against each other. This is the narrower question underneath both: one carrier, two lines, bought together, one of them quiet and one of them not.
The check that almost never finds anything
Signal at the modem. Registration state. Which generation of network it attached to. Throughput on a plain download, and round trip time to somewhere local and somewhere far.
Four minutes, and roughly nineteen times in twenty the two lines come back as twins.
My instruments say identical, the customer’s week says otherwise, and both readings are honest. Everything that separates the two lines lives somewhere my rack has no view of.
Your address had a life before you got it
A mobile line does not own its public address. The carrier lends one out of a pool when the modem attaches and takes it back when it lets go, and those addresses have been going round for years.
The one you are on this afternoon was on a phone this morning. A tablet last week. Something in a rack in another building the month before that.
Whatever the last holder did, they did it wearing that address, and the platforms kept a note. Someone watching football on the train leaves nothing behind. Someone running a heavy job at a big platform leaves a mark that outlives their hold on the address.
So the two lines were bought together while their addresses were drawn separately, out of different parts of a pool with different pasts sitting in it.
This is inherited, not manufactured. And it is invisible from my side, because both addresses route perfectly well and no field anywhere says what either of them was doing in April.
One room is not one place on the network
People assume a rack is a single location. It is one room to me. To the carrier it is a set of devices attaching to a set of cells over a set of paths back into the core.
Two modems thirty centimetres apart can land on different cells. Even on the same cell they can be given different slices of it.
So one line can sit somewhere quiet while its neighbour shares its patch with several thousand people going home at half past six.
I can watch this play out across a single evening. Two lines, same shelf, same carrier: one holds a steady rate all night and the other loses half its throughput between seven and ten, then recovers.
The second modem is fine. It happens to be standing somewhere busier, and where it is standing changes from week to week. That is why the line somebody complained about a fortnight ago often behaves impeccably by the time I look at it. The complaint was real when he made it. The congestion had moved on.
Most of the time it is the job, not the line
The one that explains most of these cases is not a property of the line at all.
The two lines were pointed at different things.
A line that has spent three months talking to one site and almost nothing else has a very specific footprint. That site has watched the address turn up over and over, at a certain rhythm, doing a certain sort of thing. It has an opinion.
A line spread thin across twenty destinations has almost no footprint anywhere. Nobody has seen enough of it to form a view.
So when a customer tells me line one works everywhere and line three keeps meeting friction, the question is no longer about the lines. It is what each one has been doing since he got it.
Nine times out of ten line three has one heavy destination in its history and line one does not. The friction only exists there. It reads as a worse line because that destination is the whole reason he bought it. Everywhere else it behaves exactly like its neighbour.
The swap test
Two questions settle most of this before anyone touches hardware.
Does it fail everywhere, or in one place? Open three unrelated sites through the line. If two are happy and the third is not, that is a destination problem wearing a line costume.
Does it fail for everything you do there, or for one account? If one account on that line walks straight through and another catches every check, the line was never the variable. An account drags its own history behind it.
Then the swap. Move line three’s job onto line one and line one’s job onto line three, and wait a day. If the trouble travels with the job, you have your answer and it cost nothing. If it stays with the line, the address is worth suspecting.
I run it before anything else now. It is free and it is the only test here that gives a clean yes or no.
What a rotation is, and what it is not
People ask for a rotation the way you would ask for a replacement part.
It drops the current address and takes another from the same pool. New address, new past, and you know precisely as little about the new past as you did about the old one. You are reaching into the same bag.
Sometimes the new draw is cleaner and the friction disappears. Sometimes it is worse. I watched a customer rotate four times trying to shake off a challenge and arrive at the far end with the same challenge, because what was being challenged was what he was doing.
Rotation resets the address. It does not reset you.
It also costs you something if you were signed in. An address that changes halfway through a session looks to a platform like somebody else picking up the account, the exact suspicion you were trying to escape.
What I can fix
Congestion, sometimes. Moving a line to another modem, another shelf, or a cell having a quieter night is a physical change and it takes about ten minutes.
Job allocation, always. If one line carries a heavy destination and another does not, reassigning which line does which job costs nothing and resolves it maybe half the time.
The address, freely. Releasing it and pulling a new one is one command and there is no charge for it.
Sick hardware, entirely. Dead radio, flapping registration, a modem that has quietly cooked itself in a warm rack: that is mine, I replace it, and it is the cleanest outcome of the lot. Also the rarest.
What I cannot
I cannot hand you an address with a known good past. The pool is the pool and the carrier decides what comes out of it and in what order.
I cannot make a platform forget what a line has already done in front of it.
I cannot separate you from your own account history, your own payment trail, or the machine the traffic is leaving from.
And I cannot tell you in advance which of the free lines on my shelf will suit you. That turns entirely on a past I have no way of reading.
“Can I have a better line” is a request for a past
Underneath, that message means a line with a cleaner history. It is a fair thing to want and nobody in this business can sell it to you. We do not pick the addresses. We have no window into what any platform thinks of them.
A seller who tells you his lines are clean is describing his intentions rather than his inventory. Ask him where he reads that from and see what comes back.
What can be promised honestly is a line nobody else is on, so that everything written on it from today forward belongs to you. The past comes as it comes. The future is the part you are buying.
Where my confidence runs out
I have no visibility into what any target thinks of any address. There is no lookup. No seller has one either, whatever their dashboard implies.
So when I tell somebody his line is fine, I mean everything I can measure is fine. That is a smaller claim than it sounds at midnight in a support reply.
A customer spent two weeks telling me one of his four lines was different. I kept sending back green readings. He was right and I was not listening. I swapped the modem eventually, mostly to end the argument, and the problem left with the old address. I never found out what was on it and I still do not know.
What I took from that is to stop using my own dashboard as a counterargument against a person who is watching the thing fail in front of him. Run the swap, rotate once, and if one rotation does not clear it then stop rotating, because by then the address is plainly not what is being judged.
If you want Singapore lines on real carrier SIMs, dedicated so that the only history written on them from today is yours, that is what I run.
Get new guides and videos first — join the Telegram channel.