Spare Resources — The endpoints of interconnections between individual networks take place
in data centers that follow a wide variety of practices to increase resilience. There, the level of redundancy and protection against typical failures is described by tiers, with specific guidelines as to what practices must be implemented for a data center to meet these levels and be certifiable as such. The Amsterdam Internet Exchange (AMS-IX), for example, has extended these available standards and further refined them into a list of 141 minimum baseline (technical design, operational, and business continuity) requirements for the data centers providing service to the exchange. While, as stipulated in these standards, it is recommended to overprovision network elements by a factor of two and create independent availability regions capable of securing network operations, there is currently an ongoing trend where providers are operating their networks at higher and higher loads (e.g., as Google is doing with their softwaredefined wide area network connecting their data centers). The “hotter” the network is operated, the fewer backup resources are available, and the higher the risk in case of failure, since backup paths/resources might not be available. Moreover, running a network at high utilization introduces a risk of overload, as we have seen, for instance, with popular applications like Twitter in their early days. Finally, adopting new technologies, such as software-defined networking (SDN, and its protocol, OpenFlow) could pose new vulnerabilities, for instance, with respect to the robustness of the SDN controller now introducing a new SPoF or the security of the OpenFlow protocol.