Case Study: Baidu
Hyperscale Performance Supporting China’s Biggest Search Engine
Baidu, Inc. was established in 2000. It has, since then, become the largest Chinese search engine. The company boasts a reach of over 629 million users in 138 countries, worldwide. It is estimated that Baidu handles around 60 million searches on a daily basis. The sheer size of the operation resulted in Baidu in being the first Chinese company to be included in the NASDAQ-100 index. This company is responsible for providing an index of over 760 million web pages, 80 million pictures, and 10 million multimedia content.
There are several hurdles faced by Baidu. The first is the capacity at which it needs to function. As Baidu continues to expand and as the demand for data volume increases, the capacity of the servers also need to grow. The second obstacle is density – a greater server density is required as this company’s activities proliferate. This increase in density usually leads to a larger energy consumption and a greater cost.
As the Internet continues to develop, consumer demands continue to grow. In order to accommodate this request, Baidu had to increase the number of data centers that it owns. This is an added expense to the TCO of the company, as well as an issue of scalability. The final impediment faced by Baidu is a need for a rapid delivery function. Currently, Baidu is installing server by server along with the operating systems and software installation. This process can take up two weeks to deploy several thousand servers.
The Inspur Rack Scale Server SR 4.5 Rackscale Architecture is equipped with the features to ensure the optimal running of Baidu. The growth in demand from Baidu’s servers could be solved with the white gold supply module that comes equipped with all the Inspur Rack Scale Server SR 4.5 products. This allows for the power supply to adapt and configure the number of modules required, thus, the power modules all operate in the vicinity of half its load. This results in the output of the servers being near its optimum efficiency point at all times. The entire power line test module conversion efficiency is almost 93.8 percent. Traditional server architectures can only produce up to 10 percent efficiency.
The Inspur Rack Scale Server SR 4.5 could also reduce the energy consumption, and as a result, the total production cost of the servers. This was achieved by the centralized power feature of Inspur Rack Scale Server SR 4.5. Here, along with the cooling system, the amount of energy required by the servers was drastically lowered.
The space issue presented by Baidu’s current servers could also be solved with the Inspur Rack Scale Server SR 4.5 design. It has been converted to a whole unit power supply cell, reducing the number of modules from 80 to just 8. The high density deployment of this Inspur server requires less space in data centers than traditional servers. It also utilizes the available space much better.
The central management and front maintenance design ensures that there is faster deployment of the servers. It also makes them easier to manage. The simplification and efficiency of the process ensure quicker delivery capabilities.
In 2011, Inspur and Baidu began their collaboration. With the aid of the Inspur Rack Scale Server SR 4.5, the two companies created a machine cabinet delivered data center equivalent to the social media giant, Facebook. The Inspur Rack Scale Server SR 4.5 was also able to garner Baidu a reputation for excellent capacity development, production, delivery, and quality control. There has also been a decrease in TCO due to the energy efficiency functions of the Inspur server.
The Inspur Rack Scale Server SR 4.5 technology allowed for the quick, smooth, and efficient services that Baidu provided for its over-half a-billion clients. The continued collaboration of this server and the company ensures that Baidu has the capacity to grow in all areas of Internet business opportunities. Inspur was also able to provide Baidu with a record-breaking 5000 nodes per day in its machine room.