Orateur
Description
This paper optimizes the configuration of large-scale data centers toward cost-effective, reliable and sustainable cloud supply chains. The problem involves placing incoming racks of servers within a data center to maximize demand coverage given space, power and cooling restrictions. We formulate an online integer optimization model to support rack placement decisions. We propose a tractable single-sample online approximation (SSOA) approach to multi-stage stochastic optimization, which approximates unknown parameters with one sample path and re-optimizes decisions dynamically. Theoretical results provide strong performance guarantees of SSOA in canonical online resource allocation problems. Computational results show that it can return near-optimal solutions and outperform certainty-equivalent benchmarks. Our algorithm has been packaged into a software solution deployed across Microsoft's data centers in collaboration with data center managers, contributing an interactive decision-making process at the human-machine interface. Using deployment data, econometric tests suggest a positive and statistically significant impact of our solution on data center performance, leading to 1--3 percentage point reduction in power stranding. The modeling and algorithmic solution developed in this paper can result in more efficient utilization of data center resources, resulting in multi-million-dollar annual cost savings and concomitant savings in greenhouse gas emissions.
