Cloud and Electricity Portfolio Optimization for Internet Data Centers under Uncertainties

Abstract

The introduction of cloud and electricity markets brings in promising economic potentials for cloud service providers (CSP) to increase profits. However, it is a challenging task to optimize the CSP’s portfolio decisions in both markets in the face of market dynamics and uncertainties. The goal of this paper is to investigate the strategic behaviors of a CSP in the context of cloud and electricity markets under uncertainties. A two-stage portfolio optimization framework is proposed. The first stage optimizes the CSP’s dynamic cloud pricing scheme, biddings in day-ahead wholesale electricity markets, and resource management. While the second stage considers the CSP’s real-time resource management and trading decisions in a cloud spot market and real-time wholesale electricity markets after capturing the actual values of cloud demands, market prices and renewable output fluctuations. An opportunity profit due to server redundant provisioning in the first stage is proposed, which can reflect the potential economic benefits in the second stage. Numerical studies are conducted to show the effectiveness of the proposed model.