What is the difference between big data centers in China and the United States?

&http://www.aliyun.com/zixun/aggregation/37954.html >nbsp;   Large data is a natural consequence of cloud computing, and it can be said that big data is part of the broader cloud. So, we can go back to the cloud computing field to explore the impact of large data on the data center site infrastructure. There are many senior data center experts who feel that if they take him to a running data center without telling any valid information, he cannot tell whether it is a "cloud data center" or "a non-cloud data center" in the face of a row of cabinets. Cloud technology is more reflected in the server running software, and the site infrastructure level for the data center to provide the basic functionality is consistent.

    But what is important here is that large data and cloud computing do have an impact on the construction of the data center site infrastructure, which is more reflected in the macro-planning dimension. The two key points of the macro planning of the data center site infrastructure are "availability level" and "Capacity scale". First of all, according to business requirements to determine the data center needs to achieve the level of availability, reflecting the construction indicators such as "in accordance with GB a level or in accordance with UPTIMETIERIII construction", of course, for large-scale data centers can have different levels of availability of modules of the combination scheme. and capacity size can be decomposed into electricity, area, refrigeration capacity of three indicators, for the new data center, refrigeration capacity can be in accordance with the IT power design, so the core indicator is the amount of electricity and area, the ratio of electricity to area is the intensity of density. Capacity scale reflects the construction indicators such as "total it 10,000 kw, power density 7kw/cabinet", of course, different modules can be set to different power densities to meet different business needs.

    from the "availability level" and "capacity scale" two aspects, large data and cloud computing on the data Center site infrastructure construction of the macro impact of the following areas:

    Data center monomer size is larger, Reduction in the number of small data centers

    Currently new data centers in Europe and America are mainly based on large cloud computing data centers to provide new IaaS, PAAs, etc. as the direction of development, China is also following this trend. By the end of 2010, the United States more than 2,000 square meters of data centers have more than 570, and the number of China's data centers, although nearly 450,000, but more than 2,000 square meters of data centers only 50, less than the United States on the same scale data center 9%. There is still a gap between the development level of China's data Center and Europe and America, and the application and deepening of large data and cloud computing will be a good opportunity to promote the development of China's data center.

    Data Center site infrastructure availability levels are no longer one-sided.

    Because there is cloud computing redundancy in the software layer, it is not cost-effective to pursue high levels on the ground infrastructure level. A typical example is Ebay,ebay from the original Tier4 level of all data centers to 20%tier4 and 80%tier2, the cost of construction and operational costs halved, still can meet the needs of business development.

    single cabinet power consumption to high density

    compared with traditional computing, large data and cloud computing mean higher power density, which is based on the standardization of IT equipment and pursues the lowest cost of operation and maintenance.Selection。 ebay has done an evaluation, at a certain amount of business (daily support volume fixed), the higher the cabinet density and save money, until the 28kw/cabinet (more than this density, air cooling can not meet the demand, the cabinet must use liquid refrigeration, will lead to a significant increase in costs). But China's situation is different, although it is also moving towards high-density, but the high degree is relatively moderate.

    The following table shows the differences between the needs and philosophies of the U.S. and China data centers and the resulting design differences. Although often in the media to see some large foreign data center exciting ultra-low Pue, but we must be aware of different domestic and foreign conditions, foreign programs can not be simply copied to China.

    We should draw on the successful building experience of European and American data centers, and combine our country's national conditions is the best way for our government and enterprise to develop, construct and layout data center. Through the above analysis, we present the following large data center design concerns to meet the needs of large data and cloud computing times:

    Availability-meet business availability requirements, differentiate between different business requirements for availability, and pursue higher availability levels without bias;

    Flexibility-can effectively support future business needs, through modular design methods, unified planning and design, step-by-step implementation. The variable characteristics of availability grade and power density are realized in a certain range by means of design technique.

    Economics-Select the lowest TCO scenario with total TCO (construction cost + 10 operational cost) assessment. Pue the lowest scheme is not necessarily the lowest TCO;

    Energy-efficient-reduce long-term operational costs by improving energy efficiency, and adopt innovative energy-saving technologies within the scope of business affordability;

    Manageability-using intelligent systems and monitoring techniques to automate operation and maintenance management to effectively support the growing size of the data center, reduce the number of unit operators and reduce the chance of human error.