The commercialization of artificial intelligence and ultra-low-latency connectivity is accelerating changes in business.
As the technology and cost-effectiveness of computing and network infrastructure continue to advance, their integration into commercial use and business operations will gain speed. It will become increasingly difficult to predict the IT infrastructure usage patterns with different type of Workload 1.
Data centers are also diversifying to align with these trends. Data centers are crucial facilities that form the fundamental physical components of IT infrastructure. With respect to investment strategies in line with business growth strategies, various choices such as form of ownership, scale (capacity), performance, location, and pricing policies are all the available options to data center’s clients.
Depending on the business workload patterns, it is also possible to have whole data centers designed for single tenant or customize area within data centers to meet the unique and varied needs of multi-tenant. For example, if a business application involves network-intensive Workload 1, small specialized data centers can be established in regions with a high concentration of consumers to maximize connectivity. If it involves compute-intensive Workload 1, options for flexible utilization of rack power can be considered. Industries subject to strict regulations regarding backup and security can utilize locations beyond a certain distance and integrated security monitoring services.
Definition and Role of Data Centers
A data center is a specially designed building that enables IT infrastructure, used by business applications, to perform its functions effectively. Data generated by business applications is stored, processed, and transformed. It can be processed within the data center or exchanged between data centers.
As data centers serve as facilities for storing and exchanging business applications and data, most companies operate data centers in accordance with the scale and nature of their businesses. However, types of data centers vary in terms of their typical size, power capacity requirements, location in the network, and purpose.
Specific area or room set up within a portion of a company's office space to operate IT infrastructure can also be considered as a type of data center. The definition and role of data centers haven't changed significantly over time, but their importance has grown. As businesses accelerate their digital transformation, higher levels of capacity, performance, availability, and reliability are required, along with architectural scalability and flexibility.
Components of Data Centers
Data centers require special levels of capacity and performance in areas such as electricity, cooling, fire suppression, security, and connectivity, which differ from those of regular buildings. Before move into the detail in each area, let's understand the key components and roles of data centers.
Electrical: Among the components of a data center, electricity is crucial, along with cooling. It supplies power to the racks (servers) in the computer room and the cooling equipment that dissipates the heat generated by the racks. In particular, modern racks require a wide range of electrical capacity, ranging from 2 kW to more than 20 kW of high power density. Therefore, the architectural structure should be capable of accommodating such capacity and allowing for expansion. Design considerations should include redundancy in the power supply lines to ensure uninterrupted power supply to the computer room in the event of issues with the main power outages due to utility grid failures, rolling blackouts, inclement weather, natural or man-made disasters, component failure, or equipment maintenance.
Mechanical: Cooling systems in data centers are equally important as electricity. With the increase in compute-intensive workloads such as AI and machine learning, the heat dissipation requirements have significantly increased. Moreover, these workloads exhibit patterns of concentrated resource utilization at specific times. As a result, investments in cooling equipment (such as BMS - Building Management System, DCIM - Data Center Information Management) to monitor and operate cooling capacity efficiently have also increased. Similar to the electrical, cooling components, plumbing and piping should be configured with redundancy to maintain temperature and humidity in the computer room without interruptions. Backup capacity, in addition to the primary capacity, should also be established to prepare for equipment issues.
Fire Suppression: Fire suppression systems in data centers are designed to protect servers using environmental friendly extinguishing agents rather than water, unlike office buildings. Smoke detection system that is constantly and actively sampling the air
the operational organization should be aware in advance, and the facility should include air-sampling-type fire detection systems to detect fires in advance by analyzing the components in the air. These systems detect even the smallest products of combustion and send out an alert as soon as these changes take place.
Security: Data centers require physical security facilities for internal and external access control and equipment transportation in loading decks. Additionally, there should be facilities to store all security data, including digital video surveillance system that works live 24/7, network-based for all recorded monitoring.
Telecommunication: Data center hosts services from multiple carriers (telecommunication companies) or organizations. MMR (Meet-me rooms) are important physical spaces located in the data center. MMR should be physically duplicated for redundancy. Data center clients use this space to interconnect or cross-connect to a single or multiple carriers (for redundancy) and to exchange information, which can be transmitted to individual computers via the Internet, without incurring local-loop fees.
Data Centers Scalability and Flexibility
As artificial intelligence and network technology evolve, more commercial use cases are emerging. It infrastructure is a major investment for the firm. They are paying attention to characteristics of modern workloads, aiming to maximize scalability and flexibility on their IT spending. While the cloud is a significant solution for this, it is not the only answer to transfer all IT infrastructure to a cloud environment. It is essential to consider the appropriate combination and balance of on-premises and cloud environments and determine the data center that suits company business.
The content is protected by copyright law and is the intellectual property of Daou Technologies. Unauthorized secondary processing and commercial use of this content are prohibited without prior consent."
Featured images via gettyimages.
References
1. 워크로드(workload) : 애플리케이션과 데이터 또는 애플리케이션과 데이터가 동작하는데 필요한 컴퓨팅(메모리, CPU, 스토리지)과 네트워크 자원의 작업량.
2. 네트워크 집중 워크로드(network intensive workload) : 결제(커머스), 거래(금융), 스트리밍(미디어) 등
3. 컴퓨팅 집중 워크로드(compute intensive workload) : 분석/통계, 머신러닝(인공지능), 그래픽 시뮬레이션(메타버스, 게임) 등
