Masters Thesis

LB-MAP: load-balanced middlebox assignment in a policy-driven data center

This thesis studies the Load-Balanced Middlebox Assignment Problem (LBMAP), which minimizes the communication energy cost of VM pairs while satisfying their policy requirement as well as the capacity constraint that the middlebox placed on the switches. The thesis shows that LB-MAP is equivalent to the classic minimum cost flow problem (MCF), which can be solved optimally and efficiently. A suite of efficient heuristic algorithms is also designed based on different criteria viz. VM-Based, MBBased, and VM+MB-Based. Via extensive simulations, it has been shown that all the heuristic algorithms perform close to the optimal minimum cost flow algorithm, while VM+MB-Based performs best among all the heuristic algorithms. This thesis is the first work that addresses the energy cost minimization for VM communications as well as loadbalancing for middleboxes in policy-driven data centers.