A Dynamic GA-based Flow Scheduling for Load Balancing in Fat-Tree Networks

  • 梁 文宣

Student thesis: Master's Thesis


Modern data center networks for fat-tree topology usually adopt multi-rooted hierarchical tree structure in order to achieve multiple paths capability and increase bisection bandwidth However the performance of data center networks heavily depend on routing protocols Traditional routing protocols are not suitable for modern data center topology because of the lack of multiple paths routing support Another important issue in data center networks is load-balancing Due to some static routing protocols limitation that could lead to some links are overload or underload utilization This situation could largely reduce the the performance of data center networks For these reasons we present a GA-based dynamic load-balancing routing algorithm which is heuristic and with centralized scheduling technique The GA-based routing algorithm mainly use genetic algorithm to search the optimization solutions We implement our algorithm in OpenFlow controller Ryu and Mininet emulator which is based on Software Defined Networking (SDN) architecture The evaluation results show that our algorithm can effectively achieve load-balancing and increase bisection bandwidth
Date of Award2016 Sep 6
Original languageEnglish
SupervisorSun-Yuan Hsieh (Supervisor)

Cite this