Data center are critical for the functioning of the digital economy.

These facilities are increasing in size, capacity, range and complexity, and the demand for network availability has never been higher. Up to 80 percent of the data center facility lifecycle TCO is across operations and management. So what role can AI play in all of this?

Join Sanjay Sainani, Global SVP & CTO of Huawei's Data Center Business, in an on-demand webinar to hear Huawei showcase the AI-enabled autonomous data center and discover how this could potentially introduce a new dimension to intelligent energy optimization, predictive maintenance and smart resource utilization.

Who is running and operating the data centers right now?

Data centers are multi-million dollar capital investments, requiring significant planning, operational costs, and technical maintenance. The effort required to maintain and operate data centers on a 24 hour, seven day schedule is immense. This effort requires a lot of manual input from human factors. In fact, Sainani highlights how the reliance on human factors and (sometimes) disconnects between planned PUE and actual PUE drive up the costs of data centers.

Like any business, data center operators are constantly looking to improve resource entitlements, which drives up the return on investment.

So what role can AI play here?

While it is still a new science, AI can contribute to data centers by monitoring the assets on a continuous cycle. This means tons of data is being accumulated every day. This data can be captured, re-arranged and analyzed through algorithms and AI-based neural networks. This has the potential to provide significantly better qualitative understandings of data center processes and operations.

Furthermore, combining this data with external sources only expands on the potential insights that can be gained from AI-integrated data centers. AI platforms have the potential to apply actionable insights that facilitate overall improvements to data center operations. AI and neural networks can often compute faster, more efficiently and with far more accuracy than humans, in specific areas. This will allow data centers to be driven by a data-based, best practice approach.

How are Huawei using AI in their data centers?

Huawei have reasonable experience in incorporating AI based platforms to take a look at work and operations across data center facility spaces. All the information Huawei retrieves from AI in data centers is visualized, meaning that the information is easier for technicians and other professionals to digest.

This also means data center maintenance can be largely autonomous - PUE can be optimized using AI enabled insights to manage and improve efficiency of resources being used with the data center. Many factors that contribute to data center operations change day by day, or even hour; this consistent change presents a unique problem for data center operators. AI enables better tracking of these variants in system operations and collate the information into actionable insights.

This has also allowed Huawei to determine what factors can be changed to improve operational efficiency. It has also removed human components out of the data center energy efficiency loop, ensuring safe and reliable maintenance. The human component can be integrated into the process, adding multiple layers of authentication to ensure only the best solutions are being actioned. It has also added a higher degree of prediction estimation, allowing AI to continuously improve energy efficiency in data center operations.

Unlocking value of AI

AI cooling features have seen PUE go down by 8-15 percent depending on the geographical location. The inspection insights provided by AI technologies have also reduced manual workload by 90 percent, a significant and extremely beneficial improvement. The reliable 24/7 monitoring poses significant benefits to both operators and customers.

These AI platforms also allow issues such as leaks and other anomalies to be visualized. This visualization also allows platforms and software to be trained to identify what type of leaks are occurring and how to raise relevant alerts to technicians and data center operators. Furthermore, this data can be used to evaluate and determine best practices, possibly impacting future evolutionary growth within data center construction.

Predicting these faults within various systems allows insights to be mapped and compared across regions, also impacting how data centers are planned for and built. Visualizing resources becomes significantly easier too, and the combined planning tools means operators have a better understanding of how resources are utilized in data centers.