Modeling and Evaluation of a Data Center Sovereignty with Software Failures

Technology Sovereignty is the capacity of a state to provide a technology to its constituents by developing or out-sourcing the technology without causing a dangerous dependency on a particular contributor. There are several laws to monitor and govern networks that ensure the privacy and security of consumers. However, they do not control the infrastructure or hardware dependency associated with these networks. A robust end-to-end network requires reliable hardware, software, and network architecture to handle multiple failures, increased users, and increased traffic demands. To handle multiple failures efficiently, there must be little to no dependency on the manufacturers involved. Therefore, the robustness of a network requires an adequate degree of sovereignty. In this study, Data Center Networks (DCN) are considered to understand the impact of choosing the hardware and software manufacturers appropriately on the DCN robustness. The unavailability of a component from a manufacturer or the presence of a software bug in the switches of a DCN can affect several flows, leading to multiple failures and huge losses of data. Thus, it is rational to find the best combination of manufacturers to create a network with the least dependency on the manufacturer(s). In this research, the sovereignty of a DCN is evaluated by considering different multiple-failure scenarios based on hardware and software manufacturers' reliability. Design guidelines for DCN operators are provided based on the findings of the analyses.