*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Optimizing Network Resource Allocation for Software-Defined Data Center Network
Committee:
Dr. Blough, Advisor
Dr. Owen, Chair
Dr. Y. Chang
Abstract:
The objective of the proposed research is to optimize network resource allocation for the software-defined data center network (DCN). Software-defined networking (SDN) has been a popular candidate for DCN since its emergence. However, with the huge traffic demands in data center, the crucial challenge of wisely allocating network resource to different users arises in DCN. Besides, the unique SDN architecture brings additional resource constraints such as the switch-to-controller link capacity and switch flow table size to this problem. We propose to optimize network resource allocation in software-defined DCN, including both these two new resources and the data plane bandwidth, with novel admission control mechanism and routing path selection algorithm. For the preliminary work, we first address the switch-to-controller link capacity constraint by utilizing auxiliary connections. A queueing model of auxiliary-connection-enabled switches is developed to study the network performance with multiple switch-to-controller connections. Second, an admission control mechanism in software-defined DCN is proposed to control the flow entrance to the network when the flow table is congested. For the proposed work, an optimization problem will be formulated and solved with the objective to maximize the total network throughput while guarantee user satisfaction level at the same time, constrained by both the switch-to-controller link capacity and the flow table size. The solution to this problem also returns the optimal routing path(s) for each user. Both the single-path case and the multi-path case will be investigated. Finally, a dynamic route selection algorithm with admission control considering a new cost function will be developed to further improve SDN resource allocation.
