What O2O need is not Hadoop, but big data.

The title, but hope that users do not misunderstand, here is the need for large data technology in the current development process, is not to hinder the application of O2O, and for O2O, the most difficult than the source of data. This is presumably also a lot of nuggets electric dealers, O2O manufacturers of the distress, but also the future to seize the O2O market opportunity for a watershed.

In fact, O2O is not a new term in the field of electric business, as early as the group buying site began to appear. Baidu Encyclopedia for the definition of O2O for online to Offline (on-line offline/line to line), refers to the offline business opportunities and the Internet, let the Internet become the front line of the transaction, this concept originated from the United States.

In China, the 2013 O2O into a high-speed development phase, a large number of businesses are all the influx of Qi. Micro-letter launched the "sweep shopping" function, users can scan the code, online purchase. Suning introduced "near Suning" function, users can find stores online, pay attention to promotional information, offline experience and purchase. And the Sky Cat will be advertising and promotional links to the line under the infiltration, offline publicity "Double 11" activities, online list.

In the recently published Blue Book: China Business Development Report (2013~2014), China's e-commerce into the O2O era. E-commerce is not fictitious economy, it is inseparable with the real economy, it is the commercial application of the real economy under the support of information technology, and it is the new technology and means of business realization. In the past, the integration of electronic commerce and real economy is mainly embodied in product integration, that is, the products traded on E-commerce platform originate from entities, but they are independent of each other in operation.

The domestic big Data technology company percentage point also saw O2O future development potential, after all, under the Chinese line retail occupies the overall 90%, but the online purchase is less than 10%. To this end, a percentage point set up O2O subsidiary "Technology", located in O2O large data, to assist the traditional retail industry offline data mining and analysis. As a big data start-up, the percentage point has been focused on consumer preference data for Internet companies and has launched a large data engine. In the electric business, education, tourism, media, finance, securities, manufacturing and other fields are involved.

Reach thousands of partners, also means thousands of business data volume, which is also to promote the percentage of the large data technology on the development of the level. Han Zhiyong, vice president of operations, told reporters that the background of the data processing technology is also experienced four stages, it built a large data processing platform consists of storage and processing two levels.

The underlying infrastructure naturally involves hadoop, but it is only one component, including Distributed file Systems (HADOOPHDFS), distributed SQL Databases (MySQL), distributed NoSQL Databases (Redis, MongoDB, HBase), Distributed Message Queuing (APACHEKAFKA), distributed search engine (APACHESOLR), and Essential apachezookeeper.

In particular, percent-proud real time computing, about this part of the component architecture and data flow as shown above. The Data acquisition service pushes the collected real-time data into the message queue Kafka; The data in Kafka is consumed and processed by two processing platforms BDM CEP (Big Data Management Complex Event 處理) and Storm.

Streaming real-time computing helps customers gain imperceptible performance. Such a framework of data processing, I believe all can be applied, whether business-to-business, or O2O and so on a variety of electronic business models, so O2O demand for large data processing technology does not need to be redesigned, the technical challenges it faces is not the biggest block of its development.

Sumeng, chairman of the technology company, said that the biggest challenge for O2O should be data collection, because the current offline data collection is not high availability. What percentage points do is research into the habits of Chinese consumer users, including their buying time, buying habits, and acceptable prices, which, based on such innate advantages, can help to some extent to offline retailing.

But this is far from enough, and offline retailing also needs to look for a way, which may still be in its infancy in China. In the United States, data collection for entity sales has long been noted. When you go to Wal-Mart, it gives you the shopping cart will track your behavior and record, along the path of people walking, not only to know the user's shopping habits, but also to a certain extent, supermarket shelves, the layout of goods placed. Of course now, with the popularity of WiFi seems to be easy to do, when your supermarket internal network, your consumer behavior trajectory is so easy to get.

In the planning of China's O2O prospects, Sumeng, for example, the future of shopping malls, by virtue of offline sales can not meet consumers, but the needs of people can not just meet online, which requires a combination of points. Based on previously collected data models, algorithms, as well as the mobile end of the guide, to attract people to the shopping center, and then use the data analysis to their accurate diversion and guidance.

Alibaba has been known to have the same idea in the O2O field it has invested in this year. This is the O2O technology for the concept of the market blueprint, but the specific implementation strategy, Sumeng did not disclose. To see hundred, fancy O2O Big Data market of course more than a percentage point, this is more coveted this big cake manufacturers are looking for a breakthrough.