The idea, explained
Most long-distance internet traffic does not travel through the air. Data is converted into pulses of light and sent through fibre-optic pairs inside cables laid across the seabed. A cable system includes the wet plant under water, landing stations at the coast, and backhaul links that carry traffic inland to network interconnection points. Backhaul is the connection between an edge location and the larger network.
A cable cut does not automatically disconnect every user. Internet operators can announce or withdraw routes so traffic shifts to another available path. That protection depends on more than having two cables on a map. If both cables enter the same landing station, use the same coastal corridor, share power, or feed the same inland fibre, one incident can affect both. Physical separation therefore matters as much as the number of links.
Capacity matters too. A backup route must have enough unused headroom to absorb diverted traffic. Headroom is capacity kept available above normal demand. Without it, rerouting may preserve basic connectivity while causing congestion, slower downloads or packet loss. Repair time also matters because undersea work may require a specialized vessel, permits, suitable weather and an accurate fault location.
The OECD describes redundancy and diversity as structural resilience measures: backups provide alternatives, while diversity reduces the chance that the alternatives fail together. The World Bank likewise recommends risk-informed design, redundant connectivity, multiple backhaul routes, backup power, monitoring and recovery planning. Neither approach promises zero outages. Resilience means reducing the likelihood, reach and duration of disruption while accepting that protection must be chosen for local hazards, costs and inclusion needs.
Follow a worked example.
Consider a hypothetical island with three international paths. Cable A can carry 100 gigabits per second, Cable B 80, and Cable C 40. Normal peak traffic is 150 gigabits per second, split as A = 70, B = 55 and C = 25.
Step 1: Cable A is cut, so its 70 gigabits per second must move elsewhere. Step 2: calculate spare capacity. B has 80 − 55 = 25 gigabits per second of headroom. C has 40 − 25 = 15. Together they have 40. Step 3: compare need with headroom. The network needs 70 but can absorb only 40, leaving a 30-gigabit-per-second shortfall.
The result is not necessarily a total blackout. Operators could keep 120 gigabits per second flowing, but peak demand exceeds the remaining capacity. Congestion controls or prioritization may reduce service quality until demand falls, capacity is added or A is repaired. A stronger design might add a physically separate route with at least 30 more gigabits per second of usable emergency capacity, while also checking that its landing station, power and backhaul do not share A's failure points.
Where people get stuck
Common misconception: two cables always mean full resilience. Two nominally separate systems may share a landing station, coastal trench, power supply or inland backhaul, creating a hidden common point of failure. Another mistake is counting total advertised capacity without checking how much spare capacity is actually available during peak demand.
Now try it yourself.
1. Check the core idea
A country has two 100-gigabit-per-second cables and normally uses 160 gigabits per second in total. If one cable fails, will the second cable carry all traffic without congestion? What else should planners verify?
Reveal the answer & reasoning
No. One surviving 100-gigabit-per-second cable is 60 gigabits per second short of the 160-gigabit-per-second peak, even before allowing operational margin. Planners should measure real peak demand and usable spare capacity, then verify physical route separation, landing stations, inland backhaul, power, equipment, repair arrangements and routing procedures. A backup is useful only if it remains available during the same incident and can carry enough diverted traffic.
Why this was added
The library covers web requests, DNS and satellite routing but not the physical submarine-cable system, shared failure points, spare capacity and route diversity that underpin international connectivity.