Electronic Thesis/Dissertation


Multi-objective design methodology using decision & causal Bayesian belief networks for complex data centers focused on risk and cost Open Access

Downloadable Content

Download PDF

A multi-objective energy efficient and flexible data center design has been of interest for all in the industry that are familiar with data center costs. A survey of 416 CIO's conducted by the CIO Data center Strategies Survey, suggested that the data center operating costs are steady at 25% of the entire IT budget. Much of this cost is directly attributed to constant scaling of data centers to address the computational and network demands. Additionally, the constant demand restructure negatively affects the non-functional costs of maintaining the data center. Many of the existing data center enhancement strategies include unilaterally adding physical servers & switches, virtualizing servers (including cloud computing), and intelligently scheduling tasks. To date, there have been very few data center design frameworks that accommodate multiple decision factors to address design objectives while also forecasting for future enhancements. The objective of this research is to design a framework using decision networks that can help create and enhance cost effective data centers. The total cost is measured in relation to the energy consumed by the entire data center network. The total cost is fed back in the system for future design decisions. The decision network is developed using Netica, and the data center energy is simulated using a simulation tool called GreenCloud.

Author Language Keyword Date created Type of Work Rights statement GW Unit Degree Advisor Committee Member(s) Persistent URL