Excellence in High-Density Data Center Operations and Maintenance: More Flexibility to Empower Enterprises for AI Readiness

The rapid development of AIGC has elevated enterprises' infrastructure demands to new levels. According to the China Data Center Industry Development White Paper (2023) released by the China Telecom Digital Infrastructure Industry Research Institute, the power density requirements for single racks in the internet industry are generally high, ranging from 6-8 kW. The financial industry stands in the middle at 5-8 kW, while government and manufacturing sectors have lower requirements, usually between 3-5 kW.As AI applications continue to evolve, increasing demand for GPU computing and heterogeneous parallel computing with GPU+CPU is pushing customers to require rack densities of 20-30 kW or even higher.Along with the rise in power consumption of racks and equipment, cooling methods have shifted from traditional air cooling to diverse solutions like cold plates, immersion, and phase change. Each method has its advantages and disadvantages. Similarly, in areas like heat exchange, wastewater treatment, and backup power design, there is no one-size-fits-all approach. As power in infrastructure increases, every component becomes interconnected. With the rapidly evolving AI landscape, how can data center deployments remain flexible? How can excellent operations be achieved? This article may provide some answers.

In the AIGC era, we are on the brink of a new generation of data centersInfrastructure for the AIGC Era

Currently, in the new generation of AI training applications represented by large models, the peak power consumption of a single SXM interface H800 GPU can reach 700W. With eight accelerators installed in one server, the peak power consumption of a single server can easily surpass 5600W.Industry expectations suggest that power consumption for flagship AI acceleration GPUs will continue to rise, and the overall power consumption of AI servers will follow suit (Source: CPU.cn).Increasing rack power not only allows for multiple AI servers to be installed in one rack but also significantly improves space utilization in data centers and reduces overall total cost of ownership (TCO).

At the same time, with the noticeable rise in AI server power, it is crucial to consider various “cooling, power, and water” issues like power transformation, distribution, cooling, UPS, and backup power, along with their corresponding operational management challenges. Whether for new construction or retrofits, planning and building next-generation AI data centers is a complex, systemic issue that requires experienced professional teams.

With over a decade of experience in hyper-scale data center construction and operations in China, Chayora, through its partner EdgeConneX's global platform, can help clients quickly access more than 50 key markets worldwide, such as Malaysia and Indonesia . Chayora has recognized many pain points clients face in AI business operations and recently launched its Ingenuity High-Density Data Center build-to-suit Solution. This solution considers the high power and density needs of AI business while providing more flexible data center deployment options to improve clients' overall operational efficiency.

Currently, the Ingenuity Solution supports up to 40 kW per rack, enabling clients to deploy and utilize valuable computing resources. Based on proven market designs, it offers flexible deployment options, allowing clients to customize specific plans according to business scenarios and enterprise needs with the support of Chayora’s internationally-minded expert data center teams. For example, considering the high power density and power consumption of AI data centers, the Ingenuity Solution offers a hybrid cooling solution combining air and liquid to ensure the stable operation of high-load applications like large AI models while also reducing overall costs.

Additionally, with the support of professional design and construction teams and a robust supply chain, Chayora works closely with clients to further accelerate installation and construction speeds in data centers. This enables faster, more flexible business launches. Overall, the Ingenuity High-Density Data Center Solution helps clients balance energy efficiency, cost, and reliability.

Excellence in Operations and Maintenance: Empowering Enterprises for AI Readiness

The AIGC era also presents new challenges for data center operations. Leveraging its years of experience in operating hyper-scale data center campuses, Chayora has its own answers.

Firstly, transparency with clients is essential. From visitor access to the data center to equipment operations, Chayora maintains complete transparency with clients to ensure open communication. Flexible strategies are crucial for efficient operations. Chayora tailors operational strategies according to the specific business characteristics and needs of clients, quickly responding to demands to provide continuous and effective support.

In addition, robust risk and disaster recovery plans are critical.Various unpredictable natural disasters or human errors can cause significant losses to data centers.Chayora has comprehensive strategies and procedures, such as emergency responses, to ensure the safety of clients' businesses. To avoid or reduce human errors, Chayora also conducts regular, customized personnel training and assessments to ensure high-quality operational staff.

Monitoring data centers is indispensable.Only proactive and continuous monitoring can better prevent risks.After all, preventing accidents is the best way to avoid losses.Chayora has a dedicated 360-degree centralized management system for data center operations, equipped with intelligent technology to monitor equipment power consumption, temperature, humidity, and other factors. This helps effectively prevent accidents. At the recently held 11th Data Center Standards Conference, this management system has been granted the Data Center Technology Achievement Award issued by China Association for Engineering Construction Standardization and approved by the National Office For Science & Technology Awards of Ministry of Science and Technology. This reflects the high level of recognition that Chayora's products and services have received from the domestic industry.

The range of specialized data center management and operational services provided by Chayora ensures high-standard support for the long-term, efficient operation of data centers. This also means that users can focus more on business expansion and innovation. Currently, Chayora is operating a hyper-scale data center in Tianjin, with construction designed to fully meet the needs of high-density rack deployments, offering strong computing power to support various applications, including large models.The Tianjin data center campus also boasts excellent energy efficiency, with an average operational PUE as low as 1.25.Since its establishment, the Tianjin campus has not experienced any major incidents and has demonstrated outstanding safety and reliability in operations, earning high praise and a strong reputation from its clients.

Staying on Course Toward a Smart and Digital Future

New-generation AI technologies, represented by AIGC, have already demonstrated significant creativity and widespread commercial value across numerous application scenarios. To drive rapid AI iteration and innovation, infrastructure must align with industry developments, providing clients with more flexible deployments and superior operations to meet their business growth, cost management, and regulatory requirements. Chayora helps clients stay on course in navigating the challenges and opportunities of their business, ultimately guiding them toward a smart and digital future.

Reprinted from "IT Grape Skin"，Click to read the original article.。