The Eternal Conflict Between Network Resiliency, Latency, & Cost

Network resiliency defined as up time entails higher build and operating costs. Resiliency in the subsea cable world reflects two basic principles: good construction practice and physical diversity. Good construction includes an undersea route that minimizes damage and time to repair. In practice this means avoiding areas where there are geophysical threats. These threats include ships, debris slides, earthquakes, and strong undersea currents that erode the protecting shielding of deep sea cables. Good practices include deep burial, undersea repeater redundancy (the number of  spare amplifiers in an undersea repeater), cable armor thickness, etc. Physical diversity means putting a big distance end-to-end between the subsea network and other submarine cables. The farther apart, the less likely a common event disrupts two or more cables. In most cases this means longer undersea paths that increase the construction bill as well as planning costs. Good examples include  the Apricot cable in Southeast Asia or the two old Hibernia cables (EXA North & South. The Apricot cable takes a long detour around Indonesia to avoid the South China Sea. This adds easily over a thousand kilometers. It also means traversing very shallow continental shelf waters where ships become a big threat. So burial depth must be greater. In case of Hibernia North and South terrestrial back haul is much more extensive than cables landing nearer to New York and London. 

Moreover, diversity is an end to end proposition so the terrestrial side including the landing spots must also be diverse. Again this entails taking longer, more expensive routes that detour around the shorter, already existing fibre optic paths. In addition, operating costs increase with distance due to the need for more optical amplification nodes and power. Maintenance contracts are also more expensive due to the longer span. 

Physical diversity often entails greater latency. The first cables to connect New York City to London took the shortest paths, including landing in the same spots on the UK side near the coastal town of Bude, UK. I count nine cables landing at Bude, UK including possibly one decommissioned system. Bude shows that carriers and even OTTs will sacrifice diversity to cut their costs or minimize latency.

These nine cables are
1. GLO-1.
2. EIG.
3. Yellow.
4. Apollo.
5. Pan-European Crossing.
6. Beaufort.
7. Amitié.
8. Grace Hopper.
9. 2Africa.

Vodafone operates two Bude cable landing stations, which are near a large British spy and surveillance facility. There is also a BT CLS as well as a former Lumen structure. Landing in Cornwall minimizes the distance to London. Moreover, the entire region is a cable landing zone so approvals are fast and simple. There is also extensive back haul fibre from the CLS to London. I believe an unspoken reason is that the British government likes to tap international cables and feels that security trumps the right to privacy. So they may either pressure cable operators to land near their spy facility or make it very easy to do so. Note that encryption does partially work against tapping. It protects the data payload. But IP packet headers are not encrypted because it would undermine the scalability of BGP routing. So tapping does not prevent nefarious characters including state actors like the NSA or their British equivalents from tracking the end point IP addresses. 

Bude is essentially a high risk single point of failure for the British telecommunications networks. Carriers know it. The British government knows it. But there has been no serious effort to diversify the dry portion of these subsea networks.