Cost analysis of the hottest data center

  • Detail

Cost analysis of data center

with the increasing size of the current data center, energy consumption and the subsequent cost issues have increasingly become the focus of the industry. The depressed economy and increasingly fierce industry competition also require enterprises to significantly reduce the operation and maintenance costs of IT systems - the purpose of IT enterprises is that it is not only an indispensable basic production link for a large number of industrial and agricultural products, but also to do more with less money, and ultimately make information services as good as other services at a low price

to meet this challenge, it managers need to adopt a variety of cost control strategies to improve data center efficiency and maximize the use of existing resources to control costs. The author believes that the current cost of data center mainly comes from the following three aspects: construction cost, operation cost and maintenance cost. The following will be analyzed in detail for the reference of IT managers

construction cost: purchase on demand and use virtualization to integrate servers

Steve SAMs, vice president of IBM Global Data Center Services, publicly said earlier that enterprises should build according to existing needs, so as to save construction costs. He believes that "the energy utilization rate of the world's most efficient data center should be as close as possible to 100% (PUE close to 1), and about 60% of the data center cost is spent on physical hardware. Therefore, enterprises should follow a principle: only buy what they need, but don't think much about upgrading and expansion. This can not only save procurement costs, but also save energy consumption of these hardware."

the "sustainable it" solution proposed by Siemens in 2010 is one of the best examples of on-demand construction. Its data center in Beijing adopts the method of phased construction and step-by-step investment, striving to maximize the return on investment. This method does not deploy a large number of hardware at one time in the initial stage, but makes the scale of its data center grow with the growth of application demand. This design concept significantly reduces the use cost and management difficulty of Siemens data center, and fully improves the utilization efficiency of infrastructure

Siemens' innovative data center

however, the sharp increase in enterprise it demand has forced enterprises to face the problem of the cost of data center computer rooms. The cost of computer room construction limits the floor area of server racks. Enterprises need to find a balance between traditional low-density racks and high-density racks with high heat and power consumption. How to effectively improve the utilization of servers and other IT equipment in a limited space is a major problem faced by the data center. At present, the utilization rate of servers in most data centers is usually only 5%-15%, and the utilization rate of direct attached storage systems is usually only 20%-40%. In terms of data utilization, 70% of data in enterprise systems is almost never accessed - in other words, the efficiency of current IT systems is very low

in the past two years, centralized integration through virtualization has become a powerful way to improve the utilization of data centers and save energy

virtualization is one of the hottest keywords in the IT industry in recent years. By integrating and pooling computing resources such as servers, it can reduce hardware procurement costs, achieve flexible allocation of IT resources, and then improve system resource utilization. For example: according to the old it application, four servers run four applications respectively, and the power consumption of each server is 2kW, while the resource occupancy rate of each server is only 10%, and the total power consumption is 8kw; The introduction of virtualization technology and the deployment of four applications on more advanced servers may increase the resource occupancy of the server to 70%, while the total power consumption is only 4kw. For an enterprise that needs to purchase new equipment, using virtualization technology will have advantages in it procurement cost, energy consumption and even green environmental protection

the deployment of data center virtualization can reduce the number of servers to a certain extent and improve the utilization of servers. By reducing the floor area and maintenance cost of the data center computer room, the service quality of the application system is improved, which plays a vital role in the integration of the data center server

however, the purchase cost of servers is only a part of it total cost of ownership (TCO), which also includes server energy consumption costs, maintenance costs and service costs

operating costs: green energy-saving data centers will become the future trend

according to the U.S. Environmental Protection Agency, the energy consumption of servers will double in the next five years. According to a Gartner survey of CIOs, 70% of respondents believe that power and cooling are the biggest problems facing their data centers. The proportion of energy expenditure in IT budget has surged from 8% to 48%. In the United States, the three-year pure energy cost consumption has been equal to the equipment purchase cost. In Europe, the cost of pure energy consumption in three years has even doubled the cost of equipment acquisition. In the future, as the problems of power supply and heat dissipation become more common and prominent, energy becomes more and more scarce and expensive - how to effectively reduce energy consumption expenditure and build a green and energy-saving data center has become an urgent problem for current IT managers

in order to build a cheaper and more efficient data center, major enterprises have invested in the deployment of green data centers. Facebook, a social networking site, recently built a data center in prinville, which is known as the most energy-efficient in the world: its power efficiency (PUE) is 1.073, which means that 93% of the power entering the center can actually be used by servers. The data center adopts special power supply equipment, which can use 277v AC power supply instead of the ordinary 208v power supply, which makes the electric energy enter the building directly and then be processed and connected to the server, without the conversion from AC to DC, so as to avoid the loss of electric energy. Moreover, the cooling system of Facebook data center is 100% cooled by external air, which enters from a group of ventilators on the second floor. Through filtration, it can also pass through the "spray room" to improve humidity if necessary. The whole cooling system is located on the second floor, and the air is directly blown to the server. The larger the clamping force is, the more air duct is required, and there is no need for air conditioning, so the heat dissipation energy consumption is saved to a great extent

facebook prinville data center

another example comes from Siemens: the fresh air system of Siemens Beijing data center can filter the external cold air in winter and effectively cool the data center. At the same time, it has the effect of dust prevention and greatly reduces the energy consumption of the cooling system of the data center

HP's new data center in Fort Collins will be equipped with hot channel and cold channel containment systems. Each channel between server cabinets is equipped with cold air inlet and hot air outlet, so as to ensure the air convection circulation between cabinets and enhance the cooling effect of cabinets. The construction of hot channel and cold channel aims to save energy and reduce cooling cost by managing air flow. It is certain that this series of measures can effectively help it managers cope with the energy consumption challenges caused by the slow growth of the global cereal packaging industry due to the growth of the size of the data center

however, the size of the data center is the key factor to decide which refrigeration scheme to adopt. Steve SAMs of IBM said: "the refrigeration scheme suitable for large data centers is not necessarily suitable for small data centers, and vice versa. In small data centers, it is common to place refrigeration equipment near servers as much as possible, but in large data centers, traditional raised floors and peripheral coolers are the most effective." Enterprises should also choose solutions according to specific circumstances

maintenance cost: safety first requires a long-term strategic vision

recently, a fire in Shanghai Telecom Data Center has attracted attention. As a hub of modern communication, the security of data center is its top priority. The data center should always be equipped with fire prevention equipment and keep it in good condition. It is not allowed to use liquid fire extinguishers to extinguish the fire at will. Automatic fire alarm devices and gas fire extinguishers should be installed as required. Good grounding is another important measure to ensure the safe operation of high-rise computer rooms. The rationality of the grounding method directly affects the stable and reliable operation of the power supply in the machine room, and the grounding wire can also protect the safety of personnel and equipment. It is important to reduce costs, but we should not lose too much in this regard

in order to adapt to the growth of enterprises, as time goes by, the hardware of the data center is also facing upgrading. Many enterprises choose to postpone the replacement time in order to save the cost of purchasing new hardware equipment. But in fact, traditional hardware can no longer meet the needs of other emerging technologies such as server virtualization technology. On the other hand, the adoption of virtualization technology can also effectively reduce the complexity of the system and reduce the investment and cost of human resources for the data center

some experts said: "the cost saved by continuing to run on those old platforms is insignificant compared with the benefits brought by the replacement." Shortening the server replacement cycle can help enterprises save budgets. For example, virtualized server supply can replace a large number of old machines by purchasing a few virtual machines, thus helping companies save costs. In the long run, the update of the server can greatly improve the level of application software and server integration. The shortening of the replacement cycle does not mean the loss of funds. Enterprises can obtain faster processors and other advantages of technological progress. (end)

Copyright © 2011 JIN SHI