What Is a Data Center?
A data center is a set of complex facilities used to centrally process, store, transmit, exchange, and manage data. It consists of servers, storage devices, switches, routers, and firewalls.
Why Are Data Centers Important?
In the business field, data centers are indispensable for enterprise operations. They support mission-critical applications, such as customer relationship management, enterprise resource planning, and supply chain management systems. Data centers play a pivotal role in development of modern technologies such as digitalization, big data analytics, and the Internet of Things (IoT).
On a technical level, cloud computing is dependent on data centers. Data centers are the foundation for software as a service (SaaS), platform as a service (PaaS), and infrastructure as a service (IaaS). With the continued growth of mobile device use and deployment of more 5G networks, data centers will play an increasingly prominent role in processing and distributing massive amounts of data.
Data centers are also critical to scientific research. They provide large-scale computing for complex simulation and calculation, thereby promoting advancements of many scientific fields such as medicine, astronomy, and meteorology.
In a word, data centers have become the beating heart of modern information technologies and the engines of the global digital economy. With the continuous development of technologies, data centers will play a more crucial role.
Evolution of Data Centers
Data centers have gone through four phases: mainframe era, midrange computer era, Internet era, and the ongoing cloud era.
Evolution of data centers
- Mainframe era (1945–1971):
In this era, electron tubes and transistors were the main components of computers. These computers were large and expensive and consumed much power. They were mainly used for specialized military and research applications. Because they processed sensitive data, data centers (which were called data computer rooms) needed uninterruptible power supply (UPS) systems and precision air conditioners, which were expensive at that time.
- Midrange computer era (1971–1995):
This era witnessed the rapid development of large-scale integrated circuits. Mainframe computers and supercomputers were still the mainstream, but midrange computers and microcomputers began to emerge. There was significant improvement in the technology, performance, and cost-effectiveness of midrange computers. Against this backdrop, there was an explosive growth in small and midsize data equipment rooms. Operating systems also saw significant development in this era. AT&T's Unix operating system running on PDP-11 minicomputers and Microsoft's Windows operating system added fuel to the development of midrange computers.
- Internet era (1995–2005)
Before 1995, small and large data centers were largely individually constructed and operated. Data was only transmitted at a small scale and at low speeds. With the emergence of the Internet, data resources scattered across different locations could be integrated and distributed to every corner of the world through the Internet. Internet Data Centers (IDCs) emerged to meet data growth requirements. IDCs centrally collect and process data, and provide diverse services such as hosting, resource leasing, system maintenance, traffic analysis, load balancing, and intrusion detection. This decade has witnessed the fast development of the Internet and also IDCs.
- Cloud era (2005–present):
Data volumes are growing from terabytes (1 terabyte = 1024 gigabytes) to petabytes (1 petabyte = 1024 terabytes) and even exabytes (1 exabyte = 1024 petabytes). This is exerting huge pressure on data centers. The use of rack servers and blade servers with one or several rack units has become a major trend in hardware development. Virtualized and massive data storage provides technical assurance. Distributed and modular data centers are beginning to dominate the market.
In the era of mainframe and midrange computers, data centers were called data equipment rooms. With the data expansion and technology transformation, these data equipment rooms gradually evolved into what we call data centers today. This evolution was not only a conceptual change; the functions, standards, and scale of data equipment rooms differed greatly from those of Internet data centers and cloud data centers.
What Are the Types of Data Centers?
Data centers can be classified by ownership, service object, or capacity.
Classification by Ownership or Service Object
Data centers can be classified by ownership or service object into IDCs, enterprise data centers (EDCs), and national supercomputing centers (NSCs).
- IDCs are standard carrier-class equipment rooms built by telecom service operators by using existing Internet communication lines and bandwidth resources. Through the Internet, IDCs provide customers with comprehensive services such as server hosting, leasing, and related value-added services. They can offer large-scale, high-quality, secure, and reliable professional services for diverse applications. These services include server hosting, space leasing, network bandwidth wholesale, Application Service Provider (ASP), and Electronic Communication (EC) services. With IDCs, enterprises and public service departments do not need to build their own dedicated equipment rooms, lay expensive communication lines, or set up dedicated network engineer teams.
- EDCs are data centers that are built and owned by enterprises or organizations for their own needs. EDCs are core computing environments used by enterprises for computing, storing, and exchanging data. They can provide data processing, data access, and application support services for enterprises as well as for their customers and partners. Industries and enterprises with special requirements on information systems, data security, and confidentiality, such as banks, insurance companies, public service departments, and large enterprises, often build their own data centers.
- NSCs are supercomputing centers that are built by nations and house high-efficiency computers with petaFLOPS computing capabilities. Unlike IDCs and EDCs, NSCs are built and operated by nations and provide computing capabilities. NSCs are mainly used for the projects that require high computing capabilities, such as animation and visual effect rendering, weather prediction, and large-scale national scientific research projects. In addition, NSCs are responsible for developing efficient and trustworthy supercomputer systems and next-generation server systems.
Classification by Capacity
Data centers can be classified by capacity into mega, large, and small or midsize data centers.
- Mega data centers have more than 10,000 standard racks. They provide high-capacity and high-performance data storage and processing services for large enterprises and Internet service providers around the world. These data centers offer support in terms of data mining, machine learning, and artificial intelligence (AI) for enterprises and scientific research institutes.
- Large data centers have 3000 to 10,000 standard racks. They provide data storage and processing services for large enterprises or Internet companies.
- Small or midsize data centers have less than 3000 standard racks. They provide data storage and processing services for small and midsize enterprises.
What Is Data Center Infrastructure?
Data center infrastructure is the foundation for efficient, secure, and reliable data storage, processing, and distribution. It consists of multiple layers, such as hardware components, software systems, and physical infrastructure. Each layer plays an important role in ensuring the overall performance and stability of data centers.
- Servers: function as the computing core of data centers. Servers run applications and process data. Modern data centers usually use high-performance servers with multi-core processors to meet increasing computing requirements.
- Storage devices: refer to hardware used for data storage, including hard disk drives, solid-state drives (SSDs), and tape libraries. Storage devices are selected depending on the data access frequency, storage capacity, and performance requirements.
- Network devices: include routers, switches, and firewalls. Network devices are used for data transmission and network security protection in and outside data centers. High-speed network devices support high-speed data exchange between data centers and between data centers and the Internet.
- Operating system: functions as the software basis of a server and provides hardware resource management and process control functions. Common operating systems in data centers include Windows Server, Linux, and Unix.
- Virtualization technology: allows multiple virtual machines (VMs) to run on a single physical server. Each VM can independently run different operating systems and applications. The virtualization technology improves hardware resource utilization and simplifies data center management.
- Data center management software: monitors and manages hardware and software resources in data centers. Data Center Infrastructure Management (DCIM) systems, server management tools, and network monitoring systems are examples of this management software.
- Power supply: Data centers require stable and sufficient power supplies to ensure uninterrupted operations. UPS systems and backup generators are usually used to prevent service interruptions caused by power outages.
- Cooling system: Devices in data centers generate a large amount of heat. Cooling systems are critical for maintaining normal device operation and prolonging the service life of hardware. Cooling systems are classified into air cooling, water cooling, cold and hot aisle isolation, heat exchange, liquid cooling, and natural cooling. Google has a data center at the bottom of the ocean, and Alibaba has a data center at the bottom of Qiandaohu Lake. These data centers reduce energy consumption with water cooling.
- Security system: protects data centers from unauthorized access and diverse security threats. Physical security measures (such as access control systems and cameras) and network security measures (such as firewalls and intrusion detection systems) are used to safeguard data center security.
Cooling systems are selected based on factors such as the data center scale, locations, energy costs, and efficiency. The objective is to ensure the reliability and performance of data centers while minimizing energy consumption and maintaining the operating temperature range of devices.
What Are the Design Standards for Data Centers?
Data centers are protected by robust and reliable support infrastructure that consists of power subsystems, UPS systems, backup generators, ventilation and cooling equipment, fire extinguishing systems, and building security systems. Industry standards, developed by organizations such as the Telecommunications Industry Association (TIA) and Uptime Institute, assist in the design, construction, and maintenance of data center facilities. Uptime Institute defines four data center tiers.
- Tier 1: Tier 1 data centers provide basic availability and services, and are suited for non-key applications. These centers lack redundant and backup systems, which means services are vulnerable to interruption.
- Tier 2: Tier 2 data centers provide higher availability than Tier 1 data centers, and offer partial redundancy. These centers are applicable to some key applications, but not the most mission-critical applications, that can tolerate short downtime.
- Tier 3: Tier 3 data centers provide higher availability than Tier 2 data centers, and usually have redundant power and cooling systems. They are suitable for most business applications that require high availability but do not require extremely high stability.
- Tier 4: the highest data center tier. Tier 4 data centers provide the highest availability and stability, and have fully redundant and backup systems. These centers are ideal for the most mission-critical applications, such as financial transactions and healthcare services.
Data Center Tier |
Availability |
Bandwidth and Capacity |
Maintainability |
---|---|---|---|
Tier 1 data center |
Tier 1 data centers have relatively low availability (99.671%), and generally have no redundant power supply or network connection. As such, single points of failure (SPOFs) may occur. |
The bandwidth and capacity are relatively low. Higher requirements of applications and services cannot be met. |
Maintenance and upgrade may cause downtime, which affects service continuity. |
Tier 2 data center |
Tier 2 data centers have redundant power supplies and network connections to reduce SPOFs. However, planned maintenance may still lead to downtime. Typically, the availability of Tier 2 data centers is 99.749%. |
Tier 2 data centers provide higher bandwidth and capacity than Tier 1 data centers, and can meet the demands for medium-scale applications and services. |
Maintenance and upgrade may lead to certain downtime, which, however, are much more predictable and controllable than that of Tier 1 data centers. |
Tier 3 data center |
Tier 3 data centers have multiple redundant power systems and network connections, and therefore can run properly during planned and unplanned maintenance. The availability of Tier 3 data centers reaches 99.982%. |
Tier 3 data centers provide higher bandwidth and capacity than Tier 2 data centers. They can meet the demands for large- and medium-scale applications and services and support more users and larger data traffic. |
Maintenance and upgrade operations can be performed without affecting operations. However, when an unexpected event occurs, downtime may occur. |
Tier 4 data center |
Tier 4 data centers have highly redundant power systems and network connections, and offer multiple redundancy paths and dual fault tolerance mechanisms to achieve almost uninterrupted operations. Tier 4 data centers have up to 99.995% availability. |
Tier 4 data centers provide the highest bandwidth and capacity, and can meet the demands for large-scale applications and services. |
Tier 4 data centers can operate stably during maintenance and upgrade, unexpected downtime, or even device faults. Redundant systems take over services during interruptions to ensure continuous running of data centers, while users are almost unaware of the faults. |
- Author: Zhang Yanlin
- Updated on: 2024-08-06
- Views: 2952
- Average rating: