Large value of large data

The rise of social media, Internet of things and E-commerce is prompting companies to look at data strategies in the hope of digging up more business value from large data analysis.

The National Oceanic and Atmospheric Administration (NOAA) issued a detailed tsunami warning just 9 minutes after the March 11 earthquake in Japan. NOAA then made a computer simulation of the real-time data obtained by the ocean sensors and produced a tsunami impact model that appeared on YouTube sites.

NOAA's rapid response has benefited from its vast global network of ocean sensors. Through these sensors, which are placed on the sea and the seabed, NOAA is constantly acquiring information about the oceans worldwide and storing the information in data centers in New Jersey, USA. NOAA's data center, which stores more than 20Pb (1024Tb) of data, is one of the largest U.S. government databases.

Vasily Titov, chief scientist at the NOAA Tsunami Research Center, said afterwards, "although the early warning system has played a great value, it has not been fast enough to help residents of Japan's Sendai coast to avoid the tsunami in time." To analyze the exact trend of tsunami activity in a shorter period of time, NOAA has been trying to upgrade its ability to handle large data-a body that has a yearly it budget of $1 billion trillion.

Though expensive, NOAA is happy about it because the data is about life. There are also a number of businesses that focus on big data, such as ebay, Wal-Mart and China Mobile. Most of them want to tap into more business value from large amounts of data, and this is a matter of whether these companies will remain evergreen in the big data age.

For any business, the data is the most dazzling gem of its business crown. With the development of traditional business intelligence system in depth application, business decision has become more and more dependent on data. However, the traditional business intelligence system for the analysis of data, most of the enterprise's own information system generated by the operating data, these data are mostly standardized, structured. In fact, this data accounts for only a small percentage of the data that businesses can get-less than 15%.

Typically, enterprise data can be grouped into 3 types: structured, semi-structured, and unstructured. Among them, 85% of the data belong to a wide range of social networks, IoT, E-commerce and other unstructured data. The generation of these unstructured data is often accompanied by the emergence and application of new channels and technologies such as social networks, mobile computing and sensors. The more comprehensive the data the enterprise uses to analyze, the closer the analysis results to the real. Large data analysis means that companies can gain new insights from these new data and integrate them with the details of known business.

In its report on large data, IDC illustrates the commercial value of using large data: The biggest difference between a leader and another is the introduction of new data types. Companies that have not introduced new analytical techniques and new data types are unlikely to become leaders in their industry.

"The amount of data produced in the past 3 years is more than the amount of data in the past 40,000 years, and the advent of the big data age is beyond doubt," said Stephen Brobst, chief technology officer at Teradata, president of the Obama Council, past, a consultant to the US Presidential Commission. We are about to face a change, the new big data will become the urgent task of enterprise development, and conventional technology has been difficult to deal with petabytes of large-scale data volume. The challenge of this change is the success of the enterprise in the future development process must face. Only those enterprises that can use these new data patterns can build sustainable and important competitive advantages. ”

Wal-Mart, one of the first companies to benefit from large data, once owned the world's largest data warehouse system. By analyzing unstructured data such as consumer shopping behavior, Wal-Mart has become the retailer that best understands customers ' shopping habits and has created a classic business case for "beer and diapers". As early as 2007, Wal-Mart built a large data center with a storage capacity of over 4Pb. The economist said in a 2010 report that Wal-Mart has 167 times times as much data as the Library of Congress.

China Mobile Group Shanxi Co., Ltd. through large data analysis, the business operation of the whole business of targeted monitoring, early warning, tracking. System in the first time automatically capture the market changes, and then the quickest way to the designated person responsible, so that he in the shortest possible time to learn about the market.

"It's important to get business information in a comprehensive way, and sometimes it can even subvert the conclusions of conventional analysis." "The Business Support System Manager of China Mobile Group Shanxi Co., Ltd." Wang Feng said. For example, a customer using the latest Nokia mobile phone, monthly payment on time, an average of 3 times a year call customer service, using WEP and MMS business. If you follow the traditional data analysis, it may be a very high customer satisfaction, the loss of very low probability of customers. In fact, when the collection includes Weibo, new sources such as social networking customer data, the customer's real situation may be this: customers buy the phone abroad, some of the functions in the mobile phone is not available in the country, in a fixed location of mobile phones often disconnected, MMS can not be used-his experience is very poor, is facing a loss risk.

"We are breaking the boundaries of traditional data sources and focusing more on new sources of data such as social media." Get as much customer information from a variety of channels as possible and dig more value from the data. Wang Feng said.

Through the analysis of the user behavior by the large data, the Internet enterprise started generally earlier. "We built a big data analysis platform 5 years ago. On this platform, you can combine structured and unstructured data to facilitate ebay's business innovation and profit growth. "said Oliver Ratzesberger, senior director of the ebay Analytics platform.

Now, the ebay analytics platform handles up to 100PB of data a day, surpassing the daily amount of data processing on the Nasdaq exchange. In order to accurately analyze the user's shopping behavior, ebay has defined more than 500 types of data to conduct a tracking analysis of customer behavior.

In the early days, changes to every feature on ebay's web page were usually determined by the product manager who knew the feature very well, based primarily on the product manager's personal experience. Through the analysis of the user's behavior data, any modification of the function on the webpage is decided by the user. "Whenever there is a good idea or idea, we will select a certain range of users on the site to test." By analyzing the behavior of these users, see if the idea brings the desired results. Oliver Ratzesberger said.

More significant changes are reflected in advertising costs. Ebay has been spending a lot of money on internet advertising, by buying some keywords from web searches and bringing potential customers to ebay. To measure the input and output of these keyword ads, ebay has built a completely closed optimization system. With this system, you can accurately calculate the return on investment that each keyword brings to ebay. Through the optimization of advertising, since 2007, EBay sales of advertising costs have been reduced by 99%, the top sellers accounted for the percentage of total sales rose to 32%.

The benefits of big data from Wal-Mart, ebay and other leading companies have undoubtedly played a role. IBM's recent "critical Revelation of CIOs in the global CIO survey" points out that 83% of CIOs have long term plans covering business intelligence and analytics, and CIOs are starting to focus more on data than apps. The findings of the Itvalue community also show that 57% of Chinese CIOs are more focused on data than on apps.

All this is really good for the popularization of large data. On the one hand, the popularity of business intelligence, so that enterprises have a full understanding of the importance of data; On the other hand, the rise of new applications such as social media, electronic commerce and Internet of Things has broken the wall of the original value chain of the enterprise, and analyzed the data of each link of the original value chain They need to use big data strategies to break data boundaries and get a panoramic view of a more comprehensive operating and operating environment.

Since the big data is about Enterprise IQ, the ability to harness large data naturally becomes the core competence of the enterprise. This capability will help companies find the best models to support business decisions and ensure that they make close to optimal business decisions.

However, the ability to harness large data is not readily available. Although it is not difficult to obtain data from new applications such as social media, E-commerce, and Internet of things, traditional business intelligence systems and analysis software are often helpless in the face of large data, such as video, pictures, text and other unstructured data, and lack of effective analytical tools and methods. It has also found a broader profit growth point for large data suppliers.

The company's revenues in the 1th quarter of 2011 grew 18% from the same period in 2010, Teradata. In the view of the company's president and chief executive, Mike Koehler, the strong performance of the big data demand market is important for revenue growth, "and many companies face the challenge of managing and extracting data value from data that continues to grow exponentially." In addition, new data elements from sources such as network interaction, social media, mobile computing and sensors provide new opportunities for enterprises to use analytics to drive innovation and win competition. This is an extremely high requirement for the scalability and management complexity of the data warehouse, which is what we are good at. ”

In order to enhance the advantage in the field of large data analysis, Teradata also acquired Aster Data Company to enhance the ability of its unconventional analysis, break the limit of SQL analysis, and help the enterprise to get more value from the whole data.

EMC is another IT company that has big data as an important development strategy. 2011 EMC World, Big data and cloud computing became the most frequently spoken word from EMC. EMC wants to reposition the legacy storage vendor's future strategy with the help of two of large data and cloud computing.

IBM's advantages in large data areas are more comprehensive than Teradata and EMC. IBM's strengths in hardware and software have been extended on large data solutions IBM want to provide end-to-end, overall large data solutions. And the robot "Watson" won in the human-computer war, but also as IBM for its large data analysis solution added to the example.

In addition, companies such as HP and Oracle also have significant advantages in large data areas. From the current situation, due to the high technical threshold of large data, it companies that compete in large data field are still in the field of data storage, analysis and other industries with traditional advantages.

At the same time, with the enterprise customers more and more of the supplier's overall solution delivery capacity, large data is involved in data acquisition, storage, analysis and many other technologies and applications of the collection. With the explosion of large data demand, it will accelerate the acquisition of major IT companies in this area, some of the IT companies providing a single technology solution, it is likely to be difficult to escape the fate of the acquisition.

(Responsible editor: Lu Guang)