CO2 penalty and disaster aware data center and service placement for cost minimization

Abstract

Data center placement and service placement within a data center are the two most fundamental problems concerning cloud network design. In an ideal world, data centers should be placed in such a manner that there should be no data loss or service disruption, the costs incurred are minimum possible amounts, latency is minimized and it operates in a “green” way, i.e., minimal CO2 emissions and runs on green energy. The service/content within a data center must be placed at locations that minimize content loss in case of any disaster, minimum latency for various requests, and also optimize available resources at hand. In this paper, we propose a data center and service placement problem that is aware of both CO2 emissions and content loss due to disaster(s). The objective is to minimize the total cost which includes operational and setup cost of data center, and penalty associated with CO2 emissions and expected content loss. This problem is modeled as a multi-objective optimization problem. Results show that the total cost incurred tend to increase with the number of disaster events and the setup and operational expenditures tend to increase with the number of software components. Furthermore, results also show that, for a given number of software components, CO2 penalty of the CO2 aware placement is less compared to the CO2 unaware placement.