Holistic strategies for data centre design, operation, and maintenance

点击翻译按钮获取中文摘要

The computing needs of our society are experiencing exponential growth, much of which are powered by data centers (DCs). While the services provided by DCs advance in complexity, the approaches used to establish new DCs are based on archaic legacy considerations and methodologies. For instance, cooling and power capacities are overdesigned to ensure that future expansion can be readily accommodated and there are many redundant precautions to limit downtime. Thus, the initial cost of DCs is high, and there is also substantial waste in operation and maintenance costs. Herein, we propose the development of a holistic DC design, operation and maintenance tool to help streamline the pre construction and operation phases of DC infrastructure development. At present, design approaches provide little visibility of future incurring costs. The proposed tool will, in contrast, offer significant oversight and granularity of operation and maintenance costs, enabling efficient and comprehensive design while minimizing such costs. The tool will be presented to users in the form of software which leverages several integrated models: (1) an equipment repository and associated physics model for every component, (2) a load trend model which will utilize data analysis techniques to perform proper server workload allocation, (3) an equipment wear and tear physics model, which allows for part lifecycle estimation and maintenance predictions, (4) an energy consumption and thermal model, which projects cooling requirements for IT equipment. The various models will be informed through extensive experimental work followed by exhaustive beta testing by Cinnos Mission Critical Incorporated, our industry partner.****