Masters Thesis

Traffic priority maximization in policy enabled tree based data centers

Increased internet service usage places increased demand on data centers posing new problems such as degrading performance and increased security risks. Middleboxes such as load balancers and intrusion detection systems are one way of addressing the issues facing data centers. A policy chain is a sequence of middleboxes that traffic must traverse in a specified order, policy driven data centers enforce policy chains to insure data integrity and improve network performance. In situations where a data center is impacted and receives more traffic demand than available bandwidth then a decision must be made as to which requests to satisfy. This thesis examines the maximization of traffic priority in policy driven data centers using tree topologies, while adhering to bandwidth constraints. Three heuristic algorithms are proposed to address the maximization problem. A dynamic programming approach that outperforms the heuristic algorithms is proposed when the data center is under additional constraints.