Many of the presentations at the Submarine Networks EMEA 2019 conference in London this February were about individual submarine cable systems, such as SACS between Angola and Brazil, Havfrue between Denmark and the US, and SEA-ME-WE 5 between Malaysia and France. However, we have identified an increased desire for collaboration between separate cable systems to offer more resilient and diverse end-to-end connectivity.
The largest buyers of international connectivity are the so-called "hyperscalers" or OTTs, including Amazon, Facebook, Google, and Microsoft. Although they are investing their capex and opex in international connectivity through multiple submarine cable systems, what they actually need are reliable and resilient end-to-end services linking their data centers worldwide. Point-to-point links are not enough. These companies need truly diverse, survivable mesh networks.
Only submarine cable systems with multiple diverse paths, such as Pacific Crossing 1 (PC-1) between Japan and the US or GlobeNet between the US and Brazil, can continue to operate after a cable break or equipment failure. Instead, customers requiring continuous and reliable connectivity must buy capacity over multiple competing cable systems and switch their traffic between them when a failure is detected. This can be both complex and expensive for non-telcos to create.
However, there is growing interest in proposals to develop a service layer across multiple independent international cable systems to provide greater availability, resilience, and capacity between regions of the world than a single cable system can provide. In addition to re-rerouting traffic should a single cable system fail, an end-to-end connectivity service operating over a combination of cable systems could also support dynamic load balancing and multiple combinations of latency, resilience, and price to appeal to different customer needs. However, this service layer would necessitate greater interconnectivity and cooperation between competing cable systems than currently exists. Furthermore, the fees paid to the supporting cable systems would have to reflect their participation in the scheme in addition to the actual capacity used by the service.
We believe that there is a growing market for a highly resilient data center to data center service in several markets. The North Atlantic connectivity market is the prime candidate for such a service where over a dozen submarine cable systems have been deployed, with more planned; the OTTs are heavy users of (and investors in) existing infrastructure, and the countries on both sides of the ocean have similar markets and objectives. The Trans-Pacific route is a second potential market for this type of service, partly because of the number of existing submarine cable systems and the high cost of deploying more due to the long distances involved. Both cases could provide a wider range of capacity, resilience, latency, and price options than any single cable system can.
A Europe to Asia service could use a combination of the existing submarine cables through the Red Sea and Indian Ocean, the terrestrial links through Russia, overland routes through China and Turkey, and the proposed Arctic Connect cable system. However, the variety of different political and commercial issues involved would make this a far more complex prospect.
"OTTs present valuable opportunities to wholesale carriers," GLB006-000004 (February 2018)
"Why invest in cable systems across the South Atlantic?" TE0012-000602 (July 2017)
"Google, Facebook, and Microsoft continue investing in submarine cables," TE0006-001389 (May 2017)
"Transatlantic cables race to reduce latency," TE0012-000568 (January 2016)
David James, Practice Leader, Wholesale Telecoms