Failure-aware Lifespan Performance Analysis of Network Fabric in Modular Data Centers: Toward Deployment in Canada's North

Data centers have evolved from a passive element of compute infrastructure to become an active and core part of any ICT solution. Modular data centers are a promising design approach to improve resiliency of data centers, and they can play a key role in deploying ICT infrastructure in remote and inhospitable environments with low temperatures and hydro- and wind-electric capabilities. Modular data centers can also survive even with lack of continuous physical maintenance and support. Generally, the most critical part of a data center is its network fabric that could impede the whole system even if all other components are fully functional. In this work, a complete failure analysis of modular data centers using failure models of various components including servers, switches, and links is performed using a proposed Monte-Carlo approach. This approach allows us to calculate the performance of a design along its lifespan even at the terminal stages. A class of modified Tanh-Log cumulative distribution function of failure is proposed for aforementioned components in order to achieve a better fit on the real data. In this study, the real experimental data from the lanl2005 database is used to calculate the fitting parameters of the failure cumulative distributions. For the network connectivity, various topologies, such as FatTree, BCube, MDCube, and their modified topologies are considered. The performance and also the lifespan of each topology in presence of failures in various components are studied against the topology parameters using the proposed approach. Furthermore, these topologies are compared against each other in a consistent settings in order to determine what topology could deliver a higher performance and resiliency subject to the scalability and agility requirements of a target data center design.