Amid the transition taking place in the global data centre industry, market players are having to address multiple challenges at once. These include supplying enough power to fulfil the significant demands of AI workloads and supporting higher-density servers, at the same time as keeping environmental footprints down and innovating on high-efficiency cooling methods.
The Asia-Pacific region is no exception when it comes to seeking new ways to handle this transition, which is being driven by AI, 5G and cloud technology. This is reflected in rapid growth in demand, with figures from a Cushman & Wakefield report showing that the sector added 1.3GW of new supply in the first half of the year to take operational capacity to 11.6GW.
Wu Huapeng, CEO of carrier-neutral hyperscale data centre player Chindata Group, reflects on the multiple considerations required under this transition. “AI data centres need to use AI and machine-learning capabilities to improve automated operation and optimisation, including smart monitoring, prediction, fault allocation and improving operational efficiency,” he says. “They need to make the most of natural resources, achieve high-efficiency cooling and reduce energy consumption.”
‘Total’ focus
As part of meeting these many demands, data centres need to adopt a “total solution” for their operation, says Wu. In this vein, Chindata has just launched AI Data Center Total Solution 2.0, a service that addresses a variety of requirements for handling AI workloads, including challenges involving the need for high-density cabinets and serving hyperscale facilities with diverse computing environments.
“The new solution takes the latest trends in AI workloads and equipment into account, innovating with multiple systems to achieve a comprehensive technological transformation from traditional to AI-ready data centres,” says Wu.
One of the service’s key attributes is that it involves modular, prefabricated equipment that enables customers to integrate features as and when they need them. Chindata says this approach to construction enables a hyperscale project to be completed and delivered to a customer within six months in China and eight months across the rest of the Asia-Pacific region.
“This method can quickly respond to changes in market demands in a form that’s like having toy building blocks,” says Wu.
Power and cooling
As Wu explains, almost all the power supply and cooling systems in the service are composed of modular parts that can be deployed in stages in line with needs. Its ‘X-Power’ system integrates modules covering substations, generators, medium- and low-voltage units, batteries and IT support equipment.
Meanwhile, the ‘X-Cooling’ components comprise comprehensive systems that fulfil the need for both air and liquid cooling, plus hybrid methods. Such a combination enables significant flexibility, allowing customers to adapt to differing architectural, environmental and geographical constraints.
These features also support the transition from traditional air-cooling methods to liquid cooling, says Wu, something he says will be needed in the long term to substantially boost efficiency and power-usage effectiveness, future-proofing data centres for the AI age.
For its part, Chindata is exploring the future possibilities for cold-plate liquid-cooling systems, and is collaborating with industry players and research institutes to that end.
“Air cooling has reached its performance limit and can no longer meet the cooling demands of high-density cabinets,” he says. “Large-scale experimentation on liquid cooling and validation of methods are therefore critical as inevitable steps towards widespread use in the future.”
As a further element in Chindata’s AI Data Center Total Solution 2.0, the service wraps in AI-driven operations and maintenance, enabling more automated management of systems.
In all, says Wu, the company’s approach will help to address all the latest challenges faced by data centres in this current phase of transition.
“We will continue to strategically plan advanced technologies and solutions for AI,” says Wu. “Our technology and experience enable us to take the lead in embracing the transition from the traditional to the AI-ready data centre.”
