Google's success in the field of search is undoubtedly inseparable from its advanced search algorithm, but there are some freshly known secrets: In fact, Google is more precious than its search algorithm than its search algorithm.
Google does not cost hundreds of or even hundreds of millions of dollars to purchase expensive devices, but only millions of dollars purchased cheap computers to build their infrastructure. By studying hardware costs, Google's technicians have found that purchasing some high-end servers is much higher than dozens of simple "popular" servers.
But after purchasing a cheap server, the problem is how to coordinate the operation of these hardware, ensuring that a computer's failure will not affect the entire system, and the task can be done normally, such as returning a search result or displays an ad.
To this end, Google considers home PC, after all, the collapse caused by household PCs due to software and hardware defects will happen every 3 years. With the size of Google, if there are thousands of PCs, a PC crash is normal every day. Therefore, it is best to solve this problem in an automated manner, otherwise the service will definitely appear.
To this end, Google, which attracts the world's best computer mind, has developed a large number of software tools to complete the installation of computing devices.
Google has its own file system, called "Google File System", which is optimized for processing large data, which can process 64MB of data blocks. More importantly, it can deal with disk or network failures that may occur at any time. Google's data is copied three, and stored in different places so that it is not very lost. With these measures to cope with fault, the PC can fully undertake the heavy responsibility of the Internet search service.
GOOGLE thousands of PC servers run a simplified version of Linux based on Red HAT, which has been modified for Google's special applications.
Google also designed a system that handles a large amount of data and quickly responding to queries. Google divides the entire Web into millions of pieces, with Google's technical terminology, these debriss are called Shard, which can be replicated when the system is wrong.
Google creates a vocabulary index appearing on the Web, and it also has a document server stores Google's current page.
Google Another important technological innovation in data center management is a software system that is able to smooth running on thousands of servers. Typically, the software system developed in parallel to multiple servers requires special programming tools and opportunities.
Google's programming tool is called MapReduce. In the event of an error, it automatically restores the entire program, which is critical to cut costs. Since last year, Google has begun to use MapReduce programming tools in large-scale.
In addition, Google also developed a bulk task scheduling software Global Work Queue, which can schedule for millions of operations. The software system can break the task into many smaller calculation operations and assign them to each computer.
In order to solve the problem of emergency disastrous problems, Google also prepared six rescue trains to deal with emergency events occurred in Google data centers. In addition, electricity cost is another important factor in Google Data Center Design. Due to the purchase of more cheap computing equipment, the overall power consumption will increase, and the power expenditure is also a major issue that the Google Design Data Center must consider.
