I. Introduction

Since the turn of the 21st century, the data center industry has been booming; perhaps its rapid industrialization has left insufficient time for more in-depth and meticulous theoretical research on data center systems. Even some related concepts still lack a unified definition, resulting in an era in which the industry is characterized by a mix of high-quality and low-quality offerings.

Regarding the concept of “data center,” numerous organizations and scholars both domestically and internationally offer differing definitions: some define a data center as a basic physical infrastructure; others view it as a physical space; while still others define it as an information system. In China’s current GB 50174-2017 Code for Design of Data Centers, a data center is defined as “a building or complex of buildings, or a portion thereof, that provides an operational environment for centrally located electronic information equipment, and includes the computer room, auxiliary areas, support areas, and administrative management areas.” ^[1] In the U.S. TIA-942 “Telecommunications Infrastructure Standard for Data Centers,” a data center is defined as: “a building or a portion of a building whose primary function is to house a computer room and its supporting areas.” ^[2] Meanwhile, Wikipedia defines it as a comprehensive and complex infrastructure that encompasses computer systems, associated equipment, redundant data communication links, environmental control systems, and various security devices; Baidu Baike defines it as a globally collaborative network of specialized equipment designed to transmit, accelerate, display, compute, and store data and information over the Internet’s network infrastructure.

As evident from the foregoing discussion, there is currently no universally accepted standard for the concept of a data center, and many more intricate logical relationships remain to be further analyzed and clarified. As the saying goes, “If names are not correct, words will not accord with the facts; if words do not accord with the facts, nothing can be accomplished.” Only by standardizing the concepts underlying all relevant systems and by clearly delineating their interrelationships can we distill the essence and build upon it in the planning and construction of data centers, thereby ensuring the sustained development of the data-center industry on a foundation of deep roots and solid stability.

Literally speaking, “data center” refers to a collection of physical symbols that carry specific meaning; in computer information science, data is defined as the collective term for symbols that can be input into and processed by computer systems. ^[3] “A data system is a system composed of five components—data acquisition, data storage, data updating, data circulation, and data mining—that is distributed across different hierarchical layers for storage.” ^[4] “Center” refers to the central position within a physical spatial extent or the weighted significance in a logical relationship; accordingly, the “center of data” denotes the weight assigned to a topological node within a computer data-processing system. Taking into account the specific functions of a data center, we can conclude that a data center is the topological core of a wide-area electronic information aggregation and processing system. The technical facility that supports the continuous operation of this core is known as the data center machine room.

Given the critical importance and inherent complexity of data centers, numerous relevant factors are carefully considered during the design and construction of data center facilities. Collectively, these factors constitute the data center’s environmental system, which can be hierarchically divided from the outside in into four levels: the external environment, the building environment, the facility environment, and the support-environment. The external environment encompasses the requirements for safeguarding the data center against external environmental risks; the building environment refers to the physical structure that houses the data center; the facility environment denotes the dedicated physical zone within the building where the data center is located; and the support-environment comprises the systems and resources that directly ensure the continuous operation of the data center’s business systems. Each of these environmental layers is composed of a set of system resources with specific functions. The absence or inadequacy of any single functional system or supporting condition will result in performance bottlenecks that undermine the overall effectiveness of data center operations.

Data center operations and maintenance refers to the ongoing, deliberate interventions undertaken to ensure the functionality of data business systems. Based on the aforementioned classification of data center systems, O&M can be further divided into two major categories: system O&M and environment O&M. System O&M pertains to the maintenance of data center business systems, focusing on the software and hardware components of these systems and implementing relevant standards related to data applications, data security, and data management. Environment O&M, also known as data center facility O&M, encompasses the infrastructure within all levels of the data center’s environmental systems and adheres to the corresponding operational and maintenance standards for data center infrastructure. Consequently, data center O&M comprises two distinct hierarchical levels in terms of scope of work; from a data security perspective, these two levels should be mutually isolated. Therefore, when designing data center facility plans and O&M strategies, thorough consideration must be given to appropriate zoning and security measures.

(2) Data Center System Partitioning

Since the inception of data centers, the number of application systems housed within them has continued to grow, now encompassing dozens of functional systems across several major categories. Moreover, driven by the evolving needs of the data center industry, new application systems are constantly being added, resulting in a highly fragmented and sprawling system architecture. Different perspectives on how to classify these systems yield varying methodologies and outcomes. At present, the industry generally adheres to the data center engineering classification approach, which groups application systems according to their original domain-specific attributes. This practice stems from the longstanding dominance of data center engineering firms in system planning, where their solutions are typically assembled by integrating offerings from individual system vendors. While this classification method is suitable for data center system engineering projects, it lacks the rigor required for comprehensive data center system planning and may hinder the further development of the data center industry.

The evolution of the GB 50174 standard demonstrates that China’s industry specifications have advanced from construction standards for computer rooms to construction standards for data center infrastructure. According to a research report published in the 2022 China Data Center Development Blue Book, China’s data center industry has further evolved from focusing solely on data center infrastructure construction to encompassing integrated projects that include data services. Therefore, from the perspective of data centers and guided by the conceptual framework of “occupying space for the human silicon-based brain,” it is highly necessary to reorganize data center systems into a new architectural paradigm.

Figure 1 is a graphical representation of the logical relationships among the data center system concepts described above. From the perspective of the data center, the data center system is divided into two major categories: the data services system and the environmental systems. The data services system (hereinafter referred to as the “services system”) constitutes the application layer of the data center’s data operations and includes various electronic information systems that rely on computer hardware and software for communication, computing, storage, and other functions. Such hardware equipment is referred to as the data center’s system equipment, while the corresponding software is termed the data center’s system software. The environmental systems, on the other hand, are the multi-level environmental subsystems that ensure the normal operation of the data center and belong to the application layer of the data center’s infrastructure—what the industry commonly refers to as the scope of data center infrastructure construction. These environmental systems are independent of the data center’s services system, and all relevant domestic and international codes and standards set forth basic requirements specifically for the four environmental systems of the data center.
 

The following discussion examines the composition of environmental systems at each layer of data centers and addresses common concerns currently of public interest, with the hope that these issues will receive due attention and be improved as the data center industry continues to evolve.

(1) Data Center Perimeter Environment

The external environment constitutes the data center’s primary environmental system and represents the foremost consideration in data center design and construction. It encompasses both physical and strategic environmental requirements, and from either perspective, it is a critical factor in determining the long-term relevance and value of the data center.

At the physical-environment level, the considerations for site selection as specified in the relevant codes and standards primarily encompass objective conditions such as the geological environment, atmospheric environment, energy supply, transportation, logistics, and disaster prevention. Moreover, depending on the project’s classification, the corresponding technical specifications set out different reference values for these factors. All of these conditions are essential prerequisites for ensuring the long-term safe operation of data centers under objective circumstances. However, in the future, a data center will function for society much like the human brain: if it is compromised, the resulting consequences could be catastrophic. This aligns with the principles outlined in China’s relevant codes and standards for determining the classification levels of data centers. Therefore, if we elevate the site-selection requirements for data rooms stipulated in the existing codes and standards to the broader external-environmental system of the data center, then the planning and deployment of data-center construction must also be conducted with rigorous strategic foresight at the macro-environmental level.

The so-called strategic environmental layout encompasses two aspects: first, a strategic arrangement made in light of the strategic needs of national economic development, taking into account factors such as geographic scope and energy allocation; second, strategic protection measures against deliberate sabotage targeting data centers. The former falls within the purview of national-level planning, while the latter is the responsibility of the project developer.

Security protection for data center environments involves implementing appropriate security technologies at every level of the data center infrastructure; however, only the safeguards deployed in the perimeter environment can be classified as strategic defense. From a strategic perspective, the most fundamental protective measure a data center can adopt in its perimeter environment is concealment: if the facility remains unknown, it is less likely to become the target of a devastating external attack. Particularly in today’s era of rapid advancement in cutting-edge technologies, for critical infrastructure such as data centers, it is not enough to simply implement comprehensive security measures—strategic planning and proactive defense strategies are equally essential. Years ago, certain countries even invested heavily in locating their nation’s most important data centers in obscure, hard-to-reach settings such as caves, underground facilities, lakes, and oceans. Beyond considerations like equipment cooling and energy efficiency, such choices often serve a far greater purpose in ensuring strategic security than they do in meeting other operational needs.

At present, in the development of China’s data center industry, promotional efforts for a data center often span its entire lifecycle—and it is not uncommon to see elaborate project-name signage mounted on the building’s exterior. With the exception of certain classified projects, much of the relevant information about data centers can be readily obtained; for some equipment vendors and system integrators, this information is even completely transparent. How, then, can a data center that lacks robust confidentiality safeguards be entrusted with the critical role of serving as the nation’s information hub? Therefore, in future data-center construction—particularly for national-level or large- and medium-sized data centers—it is imperative to strengthen strategic security awareness and strive to defend against both overt threats from external sources and covert attacks from within, so as to meet society’s growing reliance on data security going forward.

(2) Data Center Building Environment

The data center building environment constitutes the secondary environment of a data center and serves as its foundational infrastructure platform, encompassing the building structure, energy supply systems, utility routing, logistics corridors, and various protective systems designed to mitigate external environmental disturbances. The rationality of this structural design significantly influences the operational efficiency of the data center throughout its entire lifecycle. Currently, with the exception of specialized data centers, most data centers are retrofitted within existing general-purpose buildings. Even in the case of specialized data centers, due to the constraints imposed by conventional building structures, customizations are typically limited to specific parameters such as dedicated room zoning, ceiling heights, and structural load capacities, all tailored to meet the requirements of the system design. This approach merely narrows the scope of modifications required for the tertiary environment, without achieving true integration with the functional requirements of the data center environment. Consequently, substantial amounts of auxiliary construction materials are still needed during the development of the data center environment to compensate for deficiencies in the building systems.

First, the introduction of these auxiliary materials causes a sharp increase in entropy within the data center environment, complicating its external structural configuration and contaminating the system operations and maintenance environment. Moreover, the lifecycle of these auxiliary materials is shorter than that of the data center building itself, forcing data centers to take substantial risks during operation to replace them. In addition, the considerable self-weight of these materials imposes additional live loads on the data center structure, thereby reducing the building’s load-bearing usage effectiveness (LBUE). On occasion, in order to improve the physical performance metrics of the data center environment, it becomes necessary to raise the overall performance standards of the building envelope as well—this not only drives up construction and operational costs but also generates large amounts of ineffective space throughout the building, thereby lowering the space usage effectiveness (SUE) of the data center.

Therefore, to achieve true low-carbon and environmentally friendly operations, future data centers must adopt more refined planning and design at the building-environment level, challenge conventional architectural paradigms, and truly embrace the principles of “designing for function” and “better to undersupply than oversupply,” thereby institutionalizing the specialized nature of data-center buildings through their very structure.

(3) Data Center Facility Environment

A data machine room is the dedicated physical compartment within a data center’s facility, constituting the third-level environmental system of the data center. It primarily encompasses a suite of protective measures designed to mitigate electromagnetic leakage, electromagnetic interference, electrostatic discharge, temperature fluctuations, dust accumulation, fire hazards, water damage, glare, pest and rodent infestation, and theft—among other risks. In essence, it functions as the data center’s “protective shell,” much like the human skull, providing a well-structured and secure enclosure for the “brain” and ensuring its proper management. By contrast, today’s data machine rooms often fall short in this regard. Given the extent to which modern society relies on data and information, future data centers will inevitably serve as the central nervous system of human civilization. Only by designing and constructing data machine rooms with the same care and rigor as a human skull—maximizing the removal of low-energy, non-functional materials to reduce entropy within the space—can we create a safer operating environment for tomorrow’s data-driven systems.

Modern data center systems represent a cross-disciplinary, cross-functional, and cross-domain system integration. However, this should not merely involve the simple assembly of individual functional subsystems; rather, it requires decomposing each original subsystem to extract its essential functions, eliminating the inherent shortcomings of the original modules, consolidating and optimizing their underlying architectures, and then organically integrating these components in accordance with new system functional objectives. The result is a revitalized, dedicated data center infrastructure that delivers enhanced functionality, greater efficiency, and a longer lifecycle. In other words, if a data center comprises ten distinct functions, the resulting system should function as a unified “10,” rather than simply a collection of ten separate “1s.”

With regard to the data center facility environment, once the physical layout of the server room has been established, the various system functions required for that environment can only be realized after the necessary equipment and materials have been installed and integrated into the facility. However, while these equipment and materials perform their intended functions, they also introduce numerous adverse effects on the data center environment.

For example, the anti-static raised floor commonly used in data centers primarily serves to dissipate static electricity. In addition, its elevated installation structure can be leveraged as a secondary cold-air distribution channel to help regulate the temperature and humidity within the data center. However, several drawbacks arise during operation: first, the material is quite heavy (100 kg/m²). ² ), which invisibly reduces the live load capacity of the building floor; in addition, the material has a relatively low load-bearing capacity (concentrated load of 300–900 kg), often necessitating the installation of additional load-bearing steel frames to meet the requirements of the system equipment; its installation method also consumes valuable headroom in the data center; the material has a short service life (10–20 years), and replacing the floor can introduce significant risks to data services; the complex installation structure restricts the deployment of underfloor auxiliary systems; the overall flatness, structural integrity, and cleanliness of the floor are difficult to maintain over the long term; maintenance and operation of the underfloor auxiliary systems are labor-intensive; this gives rise to operational blind spots and hygiene dead zones beneath the floor; the material’s inherent fire resistance is limited, thereby increasing the complexity of the data center’s fire protection system; and it also drives up construction costs, among other issues.

Another example is color-coated steel wall panels, whose primary function is to decorate the wall surface; given that the material is a metal product, additional measures must also be taken to Equipotential Furthermore, the installation of wall panels imposes additional live loads on the data center building and reduces the available floor space; their complex installation process not only creates numerous hygiene dead zones but also increases the complexity of data center maintenance; the fire performance of the materials is limited; and overall, it drives up the construction costs of the data center, among other issues.

Upon closer and more in-depth examination, numerous similar system materials and equipment can be found throughout the data center environment. It is evident that the application of these functional materials often yields far more adverse effects than the intended benefits. Such negative impacts increase operational risks and maintenance costs in the day-to-day operation and management of data centers, potentially shortening the overall lifecycle of the data center. Therefore, if we can optimize the design of each auxiliary functional system at this level of the environmental infrastructure—ensuring that it delivers the required functionality while minimizing the use of such materials and equipment—we can mitigate many of these adverse effects and create a cleaner, more streamlined data center environmental system.

(4) Supply Assurance Environment for Data Centers

The data center’s support-environment infrastructure comprises the suite of auxiliary systems that directly underpin the operation of data services, serving as an essential prerequisite for their continuous functioning; it is also referred to as the data center’s auxiliary systems. This infrastructure primarily includes cabinet racks, power distribution, network cabling, air-conditioning, and operations-and-maintenance monitoring systems—all of which provide the operational foundation for data-service equipment. Unlike the other three-tier environmental systems, the primary requirement for the support-environment systems is not merely to maintain a relatively stable environment over the long term, but rather to possess the capability to make agile, real-time adjustments in response to changing operational demands of the data-center’s equipment. The more promptly and comprehensively such adjustments are implemented, the higher the energy efficiency of the data center will be.

Defining this tier of environment as a supply-ensuring environment carries two key implications. First, it means that the functions of all auxiliary systems within the supply-ensuring environment must provide uninterrupted support for the data service systems; such assurance extends not only over time but also across multiple dimensions—quality, scale, quantity, and performance levels—to maintain a robust balance between supply and demand. Second, it underscores the stringent requirements for the operation and maintenance of these auxiliary system components, aiming to prevent any single manufacturer’s issues from introducing risks into data center operations. For a long time, these auxiliary systems have been regarded as critical metrics for evaluating the overall performance of a data center facility, making equipment selection a pivotal step in data center construction. In this regard, manufacturers compete vigorously, each offering products with unique types, architectures, and performance characteristics, all striving to differentiate themselves and capture greater commercial value.

However, it is important to note that the very advantages of such differentiation in equipment selection can sometimes impose limitations on data center operations and maintenance. At the beginning of this century, integration of data center systems was relatively loose, with each equipment manufacturer focusing solely on its own product domain. In a typical data center project, a single vendor’s products would typically be deployed only within a specific subsystem. Today, driven by the “integrated data center” concept, equipment manufacturers are positioning themselves as “solution providers” and bundling multiple mission-critical system components for sale as a single package. This has created a “marriage effect” in data centers: once a particular vendor is chosen, the entire lifecycle of the data center will be locked into a complex, inextricable relationship with that vendor. Therefore, when selecting equipment for mission-critical environmental systems, the primary considerations should be safety, reliability, and interoperability, with efforts made to avoid letting pre-existing product differentiation dominate the decision-making process—since such an approach can undermine the very purpose of having a robust, integrated mission-critical infrastructure.

As the quintessential embodiment of new-quality productive forces, humanity is increasingly in urgent need of a data center that boasts greater capacity, smaller footprint, lower energy consumption, enhanced operational stability, and smarter management—thus paving the way for the intelligent spaces of our future. Consequently, it has become an imperative task to reflect on the past, reassess the strengths and weaknesses of various system and application solutions for data centers, and recalibrate the strategic direction of data center development.

References:

[1] GB 50174-2017 “Code for Design of Data Centers,” effective January 1, 2018, issued by the Ministry of Housing and Urban–Rural Development of the People’s Republic of China.

[2] TIA-942-B-2017 Standard, “Telecommunications Infrastructure Standard for Data Centers,” American National Standards Institute (ANSI), Telecommunications Industry Association (TIA), TIA Technical Engineering Committee (TR42).

[3] Baidu Baike: Review of the Scientific Encyclopedia Entry Compilation and Application Project for “Science Popularization China”

[4] Baidu Baike: Review of the “Science Popularization China” Scientific Encyclopedia Entry Compilation and Application Project