Why one cable is never enough: route diversity and the new digital corridors
Why one cable is never enough: route diversity and the new digital corridors
Why one cable is never enough: route diversity and the new digital corridors
On 5 August, an undersea fibre cable reached the Kazakh coast on the Caspian Sea, completing the hardest stage of a new route between Kazakhstan and Azerbaijan. Reports describe it as a fresh high-speed corridor between Asia and Europe, laid along the seabed between Aktau and Sumgait and financed by the telecom operators involved rather than the state.
The politics of the region are somebody else's article. The engineering point is the interesting one, and it is a principle every network owner already knows in their gut but tends to underweight when the budget is tight. Connectivity is not a link. It is having more than one.
The operators behind that cable were not short of ways to move data between the two regions. They built another one anyway, on a separate physical path, because a single route is a single thing that can fail. That instinct scales all the way down, from a seabed crossing to the two conduits entering a building, and it is worth taking seriously long before it is tested.
On 5 August, an undersea fibre cable reached the Kazakh coast on the Caspian Sea, completing the hardest stage of a new route between Kazakhstan and Azerbaijan. Reports describe it as a fresh high-speed corridor between Asia and Europe, laid along the seabed between Aktau and Sumgait and financed by the telecom operators involved rather than the state.
The politics of the region are somebody else's article. The engineering point is the interesting one, and it is a principle every network owner already knows in their gut but tends to underweight when the budget is tight. Connectivity is not a link. It is having more than one.
The operators behind that cable were not short of ways to move data between the two regions. They built another one anyway, on a separate physical path, because a single route is a single thing that can fail. That instinct scales all the way down, from a seabed crossing to the two conduits entering a building, and it is worth taking seriously long before it is tested.
On 5 August, an undersea fibre cable reached the Kazakh coast on the Caspian Sea, completing the hardest stage of a new route between Kazakhstan and Azerbaijan. Reports describe it as a fresh high-speed corridor between Asia and Europe, laid along the seabed between Aktau and Sumgait and financed by the telecom operators involved rather than the state.
The politics of the region are somebody else's article. The engineering point is the interesting one, and it is a principle every network owner already knows in their gut but tends to underweight when the budget is tight. Connectivity is not a link. It is having more than one.
The operators behind that cable were not short of ways to move data between the two regions. They built another one anyway, on a separate physical path, because a single route is a single thing that can fail. That instinct scales all the way down, from a seabed crossing to the two conduits entering a building, and it is worth taking seriously long before it is tested.
What route diversity actually means
What route diversity actually means
What route diversity actually means
Diversity is one of those words that gets nodded at and then quietly ignored, because on paper almost any network looks diverse. Two circuits, two invoices, job done. The trouble is that two circuits are only as diverse as the glass underneath them, and the glass is where the checking stops too early.
Real diversity is physical. It means two paths that do not share a conduit, a duct bank, a bridge crossing, a building entrance or a landing station. If both of your redundant circuits run through the same trench for even a few hundred metres, you do not have two routes. You have one route and a spare invoice, and the day a contractor puts a digger through that trench, both circuits go down together.
The convergence points are rarely where you expect them. Two routes can leave a site in genuinely opposite directions and still funnel through the same river crossing ten miles out, because there is only one bridge and every carrier uses it. They can land at what look like two different facilities that turn out to share one entrance chamber. They can ride separate ducts that were spliced into the same cable during a repair years ago and never mapped back. None of this shows on the circuit paperwork, which is exactly why it survives until the day it doesn't.
This is the gap between diversity on a procurement spreadsheet and diversity in the ground. Closing it takes someone tracing the actual physical path of each route end to end, asking every provider for the real trench, and flagging where they converge. It is unglamorous work, and it is usually the step that gets skipped when a deadline is close. It is also the difference between an outage that costs you one link and an outage that costs you the site.
Diversity is one of those words that gets nodded at and then quietly ignored, because on paper almost any network looks diverse. Two circuits, two invoices, job done. The trouble is that two circuits are only as diverse as the glass underneath them, and the glass is where the checking stops too early.
Real diversity is physical. It means two paths that do not share a conduit, a duct bank, a bridge crossing, a building entrance or a landing station. If both of your redundant circuits run through the same trench for even a few hundred metres, you do not have two routes. You have one route and a spare invoice, and the day a contractor puts a digger through that trench, both circuits go down together.
The convergence points are rarely where you expect them. Two routes can leave a site in genuinely opposite directions and still funnel through the same river crossing ten miles out, because there is only one bridge and every carrier uses it. They can land at what look like two different facilities that turn out to share one entrance chamber. They can ride separate ducts that were spliced into the same cable during a repair years ago and never mapped back. None of this shows on the circuit paperwork, which is exactly why it survives until the day it doesn't.
This is the gap between diversity on a procurement spreadsheet and diversity in the ground. Closing it takes someone tracing the actual physical path of each route end to end, asking every provider for the real trench, and flagging where they converge. It is unglamorous work, and it is usually the step that gets skipped when a deadline is close. It is also the difference between an outage that costs you one link and an outage that costs you the site.
Diversity is one of those words that gets nodded at and then quietly ignored, because on paper almost any network looks diverse. Two circuits, two invoices, job done. The trouble is that two circuits are only as diverse as the glass underneath them, and the glass is where the checking stops too early.
Real diversity is physical. It means two paths that do not share a conduit, a duct bank, a bridge crossing, a building entrance or a landing station. If both of your redundant circuits run through the same trench for even a few hundred metres, you do not have two routes. You have one route and a spare invoice, and the day a contractor puts a digger through that trench, both circuits go down together.
The convergence points are rarely where you expect them. Two routes can leave a site in genuinely opposite directions and still funnel through the same river crossing ten miles out, because there is only one bridge and every carrier uses it. They can land at what look like two different facilities that turn out to share one entrance chamber. They can ride separate ducts that were spliced into the same cable during a repair years ago and never mapped back. None of this shows on the circuit paperwork, which is exactly why it survives until the day it doesn't.
This is the gap between diversity on a procurement spreadsheet and diversity in the ground. Closing it takes someone tracing the actual physical path of each route end to end, asking every provider for the real trench, and flagging where they converge. It is unglamorous work, and it is usually the step that gets skipped when a deadline is close. It is also the difference between an outage that costs you one link and an outage that costs you the site.
New corridors change the shape of the map
New corridors change the shape of the map
New corridors change the shape of the map
There is a second reason new routes matter, beyond resilience. They change distance, and distance is latency.
Every new corridor rewrites part of the map. A path that used to force traffic through a long detour can suddenly be served by something shorter and more direct, and for anything latency-sensitive that is a real gain, not a rounding error. The applications driving demand right now, AI training and inference, cloud platforms, financial systems, are exactly the ones that feel latency and feel a lost route. As the officials behind the Caspian project themselves noted, the point of the new corridor is lower-latency, more resilient transmission that reduces dependence on existing paths, and to lay the groundwork for AI, cloud and data centres in the region.
That is the pattern worth watching wherever you operate. New capacity does not just add headroom. It reshapes which locations are well connected and which are quietly stranded, and it can move a site from the wrong side of the map to the right one. A data-centre location that looked marginal because its only good path ran the long way round can become viable the moment a shorter corridor opens nearby. The reverse is true too. A site chosen for its connectivity can lose part of that advantage as traffic and capacity shift to routes it does not sit on. Connectivity is not a fixed property of a place. It moves, and it is worth re-checking rather than assuming.
There is a second reason new routes matter, beyond resilience. They change distance, and distance is latency.
Every new corridor rewrites part of the map. A path that used to force traffic through a long detour can suddenly be served by something shorter and more direct, and for anything latency-sensitive that is a real gain, not a rounding error. The applications driving demand right now, AI training and inference, cloud platforms, financial systems, are exactly the ones that feel latency and feel a lost route. As the officials behind the Caspian project themselves noted, the point of the new corridor is lower-latency, more resilient transmission that reduces dependence on existing paths, and to lay the groundwork for AI, cloud and data centres in the region.
That is the pattern worth watching wherever you operate. New capacity does not just add headroom. It reshapes which locations are well connected and which are quietly stranded, and it can move a site from the wrong side of the map to the right one. A data-centre location that looked marginal because its only good path ran the long way round can become viable the moment a shorter corridor opens nearby. The reverse is true too. A site chosen for its connectivity can lose part of that advantage as traffic and capacity shift to routes it does not sit on. Connectivity is not a fixed property of a place. It moves, and it is worth re-checking rather than assuming.
There is a second reason new routes matter, beyond resilience. They change distance, and distance is latency.
Every new corridor rewrites part of the map. A path that used to force traffic through a long detour can suddenly be served by something shorter and more direct, and for anything latency-sensitive that is a real gain, not a rounding error. The applications driving demand right now, AI training and inference, cloud platforms, financial systems, are exactly the ones that feel latency and feel a lost route. As the officials behind the Caspian project themselves noted, the point of the new corridor is lower-latency, more resilient transmission that reduces dependence on existing paths, and to lay the groundwork for AI, cloud and data centres in the region.
That is the pattern worth watching wherever you operate. New capacity does not just add headroom. It reshapes which locations are well connected and which are quietly stranded, and it can move a site from the wrong side of the map to the right one. A data-centre location that looked marginal because its only good path ran the long way round can become viable the moment a shorter corridor opens nearby. The reverse is true too. A site chosen for its connectivity can lose part of that advantage as traffic and capacity shift to routes it does not sit on. Connectivity is not a fixed property of a place. It moves, and it is worth re-checking rather than assuming.
The same principle, closer to home
The same principle, closer to home
The same principle, closer to home
You do not need a seabed to run into this. The identical logic decides whether a data centre or an enterprise site is genuinely resilient.
It shows up as diverse building entrances rather than one, so a single cut at the door cannot take you dark. It shows up as separate duct routes to separate meet-me points, not two fibres in one sheath. It shows up in the outside-plant cable itself, the loose-tube backbone that carries the bulk of the count from the street, specified with the right jacket and armouring for the path it actually runs. Get those right and the redundancy you are paying for is real. Get them wrong and you have paid twice for a single point of failure with a comforting label on it.
Think about what actually keeps a data hall online. The compute is redundant, the power has two feeds and a generator, the cooling is built to lose a unit and keep running. Then the whole thing depends on fibre that enters the building at one point because a second entrance was value-engineered out early in the project. Every other system was designed to survive a single failure, and the network was not. That is a surprisingly common shape, and it is invisible until the one entrance is the thing that fails.
The common mistake is trusting the label. Two providers, two circuits, both sold as diverse, both landing through the same entrance because it was the cheapest way in. The fix is boring and it works. Ask where the cable physically runs, insist the answer is documented, and treat any shared segment as a risk to design out before it is the thing you are explaining after an outage.
You do not need a seabed to run into this. The identical logic decides whether a data centre or an enterprise site is genuinely resilient.
It shows up as diverse building entrances rather than one, so a single cut at the door cannot take you dark. It shows up as separate duct routes to separate meet-me points, not two fibres in one sheath. It shows up in the outside-plant cable itself, the loose-tube backbone that carries the bulk of the count from the street, specified with the right jacket and armouring for the path it actually runs. Get those right and the redundancy you are paying for is real. Get them wrong and you have paid twice for a single point of failure with a comforting label on it.
Think about what actually keeps a data hall online. The compute is redundant, the power has two feeds and a generator, the cooling is built to lose a unit and keep running. Then the whole thing depends on fibre that enters the building at one point because a second entrance was value-engineered out early in the project. Every other system was designed to survive a single failure, and the network was not. That is a surprisingly common shape, and it is invisible until the one entrance is the thing that fails.
The common mistake is trusting the label. Two providers, two circuits, both sold as diverse, both landing through the same entrance because it was the cheapest way in. The fix is boring and it works. Ask where the cable physically runs, insist the answer is documented, and treat any shared segment as a risk to design out before it is the thing you are explaining after an outage.
You do not need a seabed to run into this. The identical logic decides whether a data centre or an enterprise site is genuinely resilient.
It shows up as diverse building entrances rather than one, so a single cut at the door cannot take you dark. It shows up as separate duct routes to separate meet-me points, not two fibres in one sheath. It shows up in the outside-plant cable itself, the loose-tube backbone that carries the bulk of the count from the street, specified with the right jacket and armouring for the path it actually runs. Get those right and the redundancy you are paying for is real. Get them wrong and you have paid twice for a single point of failure with a comforting label on it.
Think about what actually keeps a data hall online. The compute is redundant, the power has two feeds and a generator, the cooling is built to lose a unit and keep running. Then the whole thing depends on fibre that enters the building at one point because a second entrance was value-engineered out early in the project. Every other system was designed to survive a single failure, and the network was not. That is a surprisingly common shape, and it is invisible until the one entrance is the thing that fails.
The common mistake is trusting the label. Two providers, two circuits, both sold as diverse, both landing through the same entrance because it was the cheapest way in. The fix is boring and it works. Ask where the cable physically runs, insist the answer is documented, and treat any shared segment as a risk to design out before it is the thing you are explaining after an outage.
Where Vocom sits, and the next step
Where Vocom sits, and the next step
Where Vocom sits, and the next step
Route diversity lives in the physical layer, and the physical layer is Vocom International's work over twenty-seven years in telecommunications. Vocom does not manufacture fibre. It sources and supplies it through tier 1 manufacturing partners, built on Fujikura glass, and it specifies and provisions the routes that carry it: outside-plant loose-tube backbone at the street, diverse duct and entrance design into the building, private circuits, and cross-Pacific capacity where the demand and the compute sit on opposite sides of an ocean.
New corridors will keep opening, and each one changes what a resilient, low-latency network can look like. The principle underneath them does not change. One path is a risk, whoever owns it and however good it is. Two genuinely separate paths are a network.
If you are designing for resilience, or you are not sure how diverse your current routes really are, that is the conversation worth having. Talk to Vocom about diverse routing and outside-plant supply: vocom.ai/contact-vocomai
Route diversity lives in the physical layer, and the physical layer is Vocom International's work over twenty-seven years in telecommunications. Vocom does not manufacture fibre. It sources and supplies it through tier 1 manufacturing partners, built on Fujikura glass, and it specifies and provisions the routes that carry it: outside-plant loose-tube backbone at the street, diverse duct and entrance design into the building, private circuits, and cross-Pacific capacity where the demand and the compute sit on opposite sides of an ocean.
New corridors will keep opening, and each one changes what a resilient, low-latency network can look like. The principle underneath them does not change. One path is a risk, whoever owns it and however good it is. Two genuinely separate paths are a network.
If you are designing for resilience, or you are not sure how diverse your current routes really are, that is the conversation worth having. Talk to Vocom about diverse routing and outside-plant supply: vocom.ai/contact-vocomai
Route diversity lives in the physical layer, and the physical layer is Vocom International's work over twenty-seven years in telecommunications. Vocom does not manufacture fibre. It sources and supplies it through tier 1 manufacturing partners, built on Fujikura glass, and it specifies and provisions the routes that carry it: outside-plant loose-tube backbone at the street, diverse duct and entrance design into the building, private circuits, and cross-Pacific capacity where the demand and the compute sit on opposite sides of an ocean.
New corridors will keep opening, and each one changes what a resilient, low-latency network can look like. The principle underneath them does not change. One path is a risk, whoever owns it and however good it is. Two genuinely separate paths are a network.
If you are designing for resilience, or you are not sure how diverse your current routes really are, that is the conversation worth having. Talk to Vocom about diverse routing and outside-plant supply: vocom.ai/contact-vocomai
