Three elements of large data value mining: comprehensive integration and open

There is no doubt about the broad application of large data. Corporate CEOs and CIOs talk about big data when they talk about business strategy and it model innovation. Gartner releases "Big Data drives Rapid Changes in infrastructure and 2 billion in IT spending 2016" (rapid changes in the large information-driven infrastructure 2016 related IT spending will reach $232 billion trillion, the report said, 2013 big data-driven IT spending is expected to reach $34 billion trillion, by 2016, the global total of large data expenditure will reach 232 billion U.S. dollars. Gartner stressed that large data analysis tools will be considered an essential investment in the 2014.

The writer: Sishing, vice President of Oracle Company and technical general manager of Greater China region

How to make full use of large data, mining the commercial value of large data, thereby enhancing the competitiveness of enterprises, has become a focus of attention. This is also the direction of Oracle's efforts.

A comprehensive solution will work.

At present, more and more enterprises will make the analysis result of large data as the basis for judging the future development. At the same time, traditional business prediction logic is increasingly being replaced by new large data forecasts. However, we should carefully manage the expectations of large data, as massive data can only further develop its business value only if it is effectively managed.

The most widely known definition of large data is the 3V characteristics of large data from Gartner: Huge data volumes (Volume), fast data processing (Velocity), variable data structures, and types (produced). According to this definition, you first think of the IT system has been difficult to deal with but can not ignore the unstructured data. In other words, large data not only deal with the analysis of transaction-type data, but also social media, E-commerce, decision-making support and other information are integrated in. Now, the distributed processing technology Hadoop and NoSQL have been able to store, process, analyze and mine unstructured data, but fail to provide a comprehensive solution to meet the customer's large data requirements.

In fact, large data ranges in general sense are broader, and any complex computation involving massive data and multiple data sources is a large data category, not just unstructured data. Therefore, such as telecommunications operators have a large number of users of various types of detailed data, mobile phone switch information, mobile phone in the network registration information, mobile phone billing information, mobile Internet detailed log information, user roaming information, user subscription service information and user basic service information, can be classified as big data.

Compared with the rise of cloud computing a few years ago, the way in which big data can achieve its business value may be longer. But business users are impatient, and more and more executives tend to use large data analysis as an important basis for their business decisions. In this context, we must find a comprehensive large data solution, not only to solve the problem of unstructured data processing, but also to extend the function to the storage of massive data, distributed collection and exchange of large data, real-time fast access to massive data, statistical analysis and mining and business intelligence analysis.

A typical large data solution should be a versatile platform solution that includes storage, computation, analysis and mining of structured data, storage, processing, and processing of multiple structured data, and business intelligence analysis of large data. This solution should have the following four characteristics: soft and hard integration of large data processing, fully structured data processing capabilities, large-scale memory computing capacity, ultra-high network speed access.

(Responsible editor: Fumingli)