Design and evaluation of data center network topologies

Abstract

Large-scale data centers form the core infrastructure support for the ever expanding cloud based services, thus the performance and dependability characteristics of data centers have significant impact on these services. Significant research work has been done on designing the data center network topologies in order to improve the performance of data centers. In this thesis, we first present survey of various representative DCN topologies, along with comparisons with respect to several properties in order to highlight their advantages and disadvantages. The variety of DCN topologies leads to a strong need for a standardized method for evaluating and comparing various DCN architectures. Considering the need, we designed DCNSim, a general purpose simulator that supports most well-known DCN topologies. The modular and flexible architecture of the simulator permits easy extension to support any future proposed topologies and computing metrics. With the support of DCNSim, we present evaluations and objective comparisons of the fault-tolerance characteristics of several important fixed-topology DCN architectures. Besides individual failures, we introduce fault regions to study associated failures in DCNs. We use connection-oriented metrics and network size-oriented metrics to reveal the performance of DCN topologies from different perspectives. Based on the understanding of fixed-topology architectures, we discuss flexible-topology architectures using optical technology. The existing topology construction algorithm applied in literature cannot fully utilize such model and its flexibility. Besides, reconfiguring optical DCN topology according to the changing traffic is a problem that has never been well researched. We focus on addressing these two main problems which are of key importance to fully exploit the advantages of the optical DCNs and present algorithms addressing the problems.

Date of Award	2013
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology