Data Center Sustainability: In Search of the Big Picture
Given that the demand for data centers is rising and that they already contribute up to 2% of global greenhouse gases, data center sustainability efforts are essential. Despite this, however, there has yet to be a comprehensive analysis of how the data center industry is performing on sustainability.
To help address this, researchers from Nanyang Technological University and Singapore’s Building and Construction Authority collaborated to assess the current state of data center sustainability and report on the industry’s evolving sustainability trends.
As the researchers describe in a recent article in IEEE Transactions on Sustainable Computing, they used publicly available data to assess the sustainability status of data centers. They also developed a cognitive digital twin (CDT) approach for promoting data center sustainability.
The Research
After offering an overview of critical metrics for evaluating data center sustainability, the researchers analyzed the sustainability rating scores of data center projects available in the database of Leadership in Energy and Environmental Design (LEED), the global green-building certification program.
Among their findings were that improving energy efficiency was the most straightforward approach to raising a data center’s sustainability rating. Yet they also found that some centers with high energy efficiency failed to achieve the higher-level certifications because they performed poorly on other sustainability aspects—such as efficient water usage. The researchers further found that while most centers performed well on energy metering, many nonetheless had unsatisfactory energy performance.
To shed light on emerging optimization trends in data center sustainability, the researchers also examined the progression of data center sustainability standards in Singapore. Based on their analysis, they identified several key components of future sustainable data centers, including:
- advanced physical and cyber infrastructures,
- full integration of renewable energy, and
- accurate carbon measurement and reporting.
The Cognitive Digital Twin Approach
To support the rapid transformation of sustainability in the data center industry, the researchers created the CDT architecture to bridge the gap between traditional rule- or model-based data center operations policies and fully trained AI models. The latter requires data on myriad situations, which is often prohibitive in mission-critical data centers where safe operation is the first priority. The CDT system encapsulates a digital twin engine capable of simulating the full systems states, along with a data-driven decision engine to derive optimal control policy.
To demonstrate CDT’s effectiveness, the researchers applied it to optimize chiller plant performance in a production data center in Singapore. Their results show that the CDT is capable of not only accurately predicting the plant’s energy consumption but also of reducing carbon emissions by approximately 140 metric tons per year.
Overall Findings
Based on their review and an analysis of their own work, the researchers’ findings include the following:
- Although IT equipment is a data center’s largest energy consumer, existing sustainability rating frameworks fail to offer unified metrics for evaluating IT equipment efficiency.
- Data center sustainability goes beyond energy efficiency; a center’s sustainability efforts should also consider water efficiency, environmental impacts, and circular economy.
- Renewable energy integration is important, and future sustainability rating systems should include more credits for renewable energy use to encourage the data center industry’s green transformation.
The researchers also note that sustainable data centers should use networked sensing infrastructures with advanced DCIM systems to share their performance information in greater detail. Given such a supporting infrastructure, the CDT platform could then systematically optimize data center sustainability.
Learn More
As this research effort shows, data center sustainability is an evolving practice. The work also provides many insights into how that practice can be improved—something that remains urgent as the data center industry’s importance and growth continue to rise.
To learn more and read detailed results of this sustainability research, download “Data Center Sustainability: Revisits and Outlooks,” by Zhiwei Cao, Xin Zhou, Xiangyu Wu, Zhaomeng Zhu, Tracy Liu, Jeffery Neng, and Yonggang Wen in the IEEE CS Digital Library.