Crash me if you can: Rethinking sustainable Data Center Networking from a topological perspective
Ritchie Qi, William Liu, Jairo Gutiérrez, Asad Waqar Malik · 2016
a data center is a pool of resources such as computational, storage facilities, and servers interconnected using a communications network. The Data Center Networking (DCN) holds a pivotal role, and it needs to be scalable and efficient to connect the growing number of servers so as to handle the intensive demands in cloud computing. Recently there has been a rapidly growing literature on DCNs which focuses on studying how to model and evaluate the resource provisioning and allocation algorithms for more effective and efficient resource management of cloud systems. Unfortunately there is not much work to reveal how the underlying network topological connectivity can affect the overlay DCNs performance in aspects such as energy consumption and service resilience. There is a saying that, it is not what you know i.e., the algorithm itself but who you know i.e., connectivity, which argues that people get ahead in life based on their connections, not on their skills or knowledge, and every day offers evidence of this aphorism. This case also applies to DCNs. DCNs performance is not merely a function of resource provisioning and allocation, but also it is a network-wide activity. The structure and ties that links a data center to other data centers are also critical factors. In this paper, we are paving a new developing line of research on revealing the roles of network topological characteristics to foster more sustainable DCNs in terms of resource efficiency and service resilience. We believe this research is thought-provoking, and opens a new conversation for researchers working in the area of sustainable DCNs communications and networking.