How Far Is the Data Center Industry from Complete Automation?
GDS believes AI readiness of the data center industry can be understood by classifying data centers into five levels, whether they are at the initial stage of fully manual operations or at an advanced stage with fully automated operations, much like how the US-based SAE International (formerly the Society of Automotive Engineers) evaluates the level of automation of companies in the automated driving technology sector.
“With the increasing scale and number of data centers, technical risk and personnel management is becoming more complex,” said Senior Vice President Yan Liang. “Looking forward, it will be difficult for a data center to fully meet the requirements of safe operation and ultimate energy efficiency of large-scale data center if only continuing a manual operation and management model.”
Imagine Iron Man without Jarvis. If the AI assistant wasn’t around (virtually) to take care of all the internal systems of Stark's buildings and the Iron Man suits, Iron Man would have to enter the building from the front door and push the button for the elevator on his own. The industry will need a “Jarvis” of its own to reach the fifth level - complete automation. More research and consideration will be needed in terms of data analysis, design and management, and further exploration in regions with weak voltage power infrastructures.
Given the current AI and technological capabilities, industry players have the potential of reaching Level 4 automation in the near future i.e. best-in-class data center would be able to demonstrate automated and predictive troubleshooting and analysis capabilities, handle emergencies in automated manner and deploy AI-based efficient energy management systems so that manual operations are minimized to the greatest extent possible.
GDS has more than two decades of experience and operates 70 self-built data centers in China serving more than 700 customers, including hyperscale internet and cloud service providers, financial institutions and multinational corporations. “We have invested in cutting edge technology in recent years that assists employees across the full life cycle of design, construction and data center operation,” said Liang. “Not only can such technology make data more secure via increasingly precise threat/attack detection, greatly reducing human error, and lower latency in Deep Learning. It will also continuously optimize PUE (Power Usage Effectiveness) thereby achieving automated operation and maintenance with better energy efficiency.
With each passing year, the data center industry is getting closer to what was once considered the stuff of science-fiction and fantasy.