Driven by the adoption of enterprise cloud-based technologies and consumer video applications, data centers are rapidly proliferating around the globe. Internet content providers, communications service providers, carrier-neutral providers, and many others are massively investing billions of dollars in them. Internet content providers are small in absolute numbers of data centers; they have been building, however, some of the largest and most traffic intensive ones (the so-called web-scale data centers). Furthermore they are expanding into new geographies; where the end users are and availability of cheap power and real estate govern new locations. Internet content providers are still adding capacity to data centers today to serve more bandwidth-intensive consumer and business applications.
This data center proliferation also impacts network demands with huge traffic requirements and new patterns. The underlying applications enabled by data centers depend on the optical network for transport connectivity between users and the application (north-south) and between applications (east-west) hosted in data centers. While both are growing strongly capacity wise, the increase in east-west (or machine-to-machine) traffic is driving very-high-capacity requirements for data center interconnection. Also Internet content providers have moved on to “edge” data center construction; smaller than their web-scale counterparts, edge data centers are located closer to user populations. The objective here is to reduce latency and improve application performance. These edge data centers also have a strong effect on network demands and topology.
As a result, data center operators have invested in continental scale long haul networks and in international networks. They were among the early adopters of coherent 100G technology to drive cost per bit transported down, and are pushing for new technologies (e.g. wideband subsea repeaters, new optical transceiver modules) and new network models (e.g. open subsea cable systems).