What is big data, and what is the difference between application scenarios and traditional bi?

I recommend reading this article, which may be helpful to you ~ ~ ~

Fu Lihu, director of Hualian Shang Chao Data Center, once told a story: As a large commercial supermarket in China, Beijing Hualian has tens of millions of transaction data from stores all over the country every day, and the accumulated data purchased by users every year is only over 2TB, so the demand for data analysis and application is very strong. Therefore, in 2008, Hualian introduced the BW system of SAP for data analysis, and then introduced the BO product of SAP on 20 12 to do more advanced data analysis and guide business.

But what makes Fu Lihu helpless is that it takes about 20 minutes to query the billion-dollar report with SAP's BO, and at the same time, the online four-person system will collapse ... Expensive foreign software can't solve the problem, and Fu Lihu began to look for solutions in China, so there was a connection between Haizhi BDP and Hualian.

The story of Hualian Shang Chao is not a case. Recently, the new famous brand retail, which focuses on "fast fashion", has reached a cooperation with Haizhi BDP, because the BI system of SAP is used, and the time for data collection, extraction and display is calculated in hours, which is very inefficient. For example, it takes 6-8 hours to export a report, and the data export process is often interrupted, which brings great inconvenience to the real-time analysis of data analysts. ...

Business Intelligence, English business intelligence, abbreviated as BI. This concept was first put forward by Gartner in 1996. With the entry of overseas software giants such as SAP and Oracle into China, it was once considered as a new growth blue ocean in the field of enterprise management software after ERP.

However, the cruel reality is that the failure rate of traditional BI advocated by software giants remains high. According to incomplete statistics, in the practical application of enterprises, the failure rate of business intelligence reaches 70%, which is amazing.

The death of traditional BI is not alarmist. The high implementation failure rate reflects the multiple dilemmas of traditional BI.

The first is the technical dilemma. In fact, the cases of Hualian Shang Chao and Famous Products reflect that traditional BI technologies such as ETL, data warehouse and OLAP are on the verge of elimination, because they can't solve the problem of dealing with massive data (both structured and unstructured).

Some engineers spoke on the Internet: "The original BI miners were happy to take some samples and run an R on a single machine, but now it's not working. Try the three-degree communication circle of 50 million users? "

The computing performance in the era of "small data" makes the traditional BI difficult in the Internet era. Therefore, only updating methods can bring new opportunities. Basically, all the functions of traditional BI can be replaced by corresponding big data components, and big data technology has cost advantages, and the replacement of technology is the general trend. The second is the business dilemma. As we all know, whether IT is a large enterprise in Gao Fushuai or a small and medium-sized enterprise of 20 million in China, it is an expensive IT cost for enterprises to purchase software services from SAP and Oracle, and it is impossible to expect them to complete the enterprise informatization task in China. If technology is not universal, technology will always be a game for a few people. In addition to the high cost, the traditional software delivery mode running according to the project cycle can no longer meet the rapidly changing needs of enterprises. In the process of traditional BI implementation, the first-stage projects often look good, but the new needs and projects of enterprises will become distant or unfinished.

Fortunately, there is cloud computing. The concept of Software as a Service (SaaS) completely subverts the traditional software business-paying on demand, obtaining resources online, and rapid iteration constitute the new standard cognition of software services for enterprises in the Internet era.

Traditional BI vendors have been calling for years to "help enterprises make wise business decisions". Now, apart from a bunch of report systems, some decision trees and other statistical algorithms, what's left? Traditional enterprises have introduced so many BI consultations and written so many reports. How much value has really happened? Fundamentally, in traditional BI vendors, the target audience is only the boss, and the decision-making and implementation are out of line, so they can't sink to the front line and eventually become a face-saving project, which can't produce practical value at all. The failure of traditional BI is the result of technology hollowing out caused by technology-driven business. This development for the purpose of statement is bound to be eliminated by history.

In order for the big data of enterprises to play a role, the target audience should aim at those who really do operations, do analysis and look at data in the front line of business-why is the activity of registered members of xxx APP declining today? Why do xxx goods sell more in the morning than in the afternoon? Why is the advertisement on xxx channel ineffective for a week? ..... These real business scenes that are staged all the time can't all wait for the boss to answer. In order to get real-time results when employees have ideas in their minds, data analysis tools are needed to lower the technical threshold as much as possible, greatly improve the technical performance, and simply drag and drop to show beautiful data charts, preferably taking into account both PC and mobile. Only by making good use of data analysis can business departments maximize the value of data.

Data-driven is not only the boss, but also the data should be dissolved into the blood of every ordinary employee in the enterprise, so that data-driven will not become an empty talk.