How did the big data hit the ground after the traditional industry exploded?

During the national Day, the national attractions everywhere crowded. However, a number of attractions in upset: October 1, the number of Mount Wutai decreased by nearly 30% in the same period last year, Huangshan, the total number of visitors only with the weekend level is comparable. If the holiday before the arrival of relevant data to predict the flow of people around the site, we can prepare ahead of time, wrong peak travel. In fact, before the National Day holiday, Art Dragon, the same journey and other travel sites have been based on air tickets, hotels, attractions, tickets booking volume issued by the corresponding data analysis report: Hot destinations in Sanya, the most popular tourist destination is Beijing-Big data is quietly playing a role around us.

1. Large data applications in traditional industries

1 Beer and diapers

In Wal-Mart, the United States, beer and diapers are always placed in the same area, convenient for "milk dad" shopping, and similar to the same shelves of instant noodles and ham, chewing gum and condoms. These intimate arrangements are all benefited from the supermarket supplier of customer habits of a profound interpretation, but also benefited from large data in the large business, shopping malls in the use. In the process of transforming traditional shopping malls into commercial complex, it is very important to analyze the user's preferences and behaviors by using large data. Usually the shopping mall needs external third party data to analyze the popular brand, the food taste and consumption level of the people in a certain area, meanwhile, the internal data, such as the traffic of the corresponding period, the consumption data of POS, the CCTV monitoring data, the shopping product list data and so on are also very important.

2 Balance Treasure

Balance treasure since June 2013 on the line, only in half a year to make the day behind the fund scale growth of 40 times times. By March 2014, the number of users of the balance treasure had reached 81 million, surpassing the number of active accounts in the entire Chinese stock market. Money fund assets mainly invest in short-term monetary instruments, large-scale fund investment in pursuit of high profits, but also to avoid risk. Through the support of large data, balance treasure can forecast the trend of liquidity, timely support liquidity needs, reasonable risk avoidance. Fund companies can provide the balance through Alipay cash flow situation, through the model to predict flow trends, arrange investment. But the traditional industry does not have the big data support, naturally loses such ability.

2. Large data in traditional industries landing difficult

1 related personnel scarce. People who know both business and data are not hard to find in the Internet industry. In traditional industries, the combination of data and business is often very loose. Experts with large data expertise are clearly missing from the traditional industry's past operations. Traditional industry technicians often think big data is hot, but can't answer the following questions: What data do you get? What's the use of this data? How do I use this data?

2 The accumulation of technology is insufficient. In the Internet industry, the use of a very wide range of technology, traditional industry, the technical staff have little contact. In the actual implementation of large data, such as Hadoop, Message Queuing, storm and other large data processing technology, the traditional industry is still in the preliminary understanding stage, can really apply less.

3 Cost capital consumption. Some traditional enterprises in the preparation of independent data applications, found that the cost of implementation is very large, the overall it budget expenditure increased by 1 or even 2 orders of magnitude, and whether these inputs to achieve the expected effect is not sure.

3. Ucloud Cloud Platform Targeted Solutions

For the shortage of personnel, in addition to the positive introduction of relevant personnel and internal technical personnel training, traditional enterprises can also seek external cooperation. At present, many enterprises in these areas of expertise to provide mature solutions. Ucloud, as the most professional basic cloud computing service provider, has also targeted the development of large data products UDDP (Ucloud distributed) to help users easily and quickly deal with TB or even PB-class mass of data. At present, UDDP is being used in digital marketing, data analysis, business intelligence, scientific simulation and other fields.

For the problem of insufficient technology accumulation, the large data product service provided by Ucloud can effectively help the enterprise users to reduce the technical threshold and avoid the complicated technical details. At the same time, ucloud professional and technical leaders will also provide customers with targeted training courses, and from time to practice in the implementation of technical guidance to help users quickly eliminate technical short board.

The cost of capital is simpler, and using cloud computing services is a very effective way to reduce overall costs. By using cloud services and reusing resources, traditional enterprises can reduce the cost of large data applications on a large scale. For example, a large data product using Ucloud, estimated at 10TB per day, would cost about thousands of yuan per month.

4. Ucloud large Data product system design

In the overall system framework of ucloud large data products, the lowest level is the interface of data acquisition, which is stored on HDFS by collecting data from camera, sensor, various logs and database. On top of storage, we support two distributed computing frameworks, Spark and mapreduce, for use by different custom users. It also provides column-type storage hbase.

Above the data, there will be a lot of tool support, such as Spark based Spark SQL, Spark streaming, Spark Mlib, and MapReduce, Hive, and so on. The computed data can continue to be stored back into the HDFS, or it can be exported to db or elsewhere for other purposes, creating a variety of business models to use.

In the cluster, we will dispatch the data Analysis task and make full use of the cluster resources. At the same time, the cluster monitoring, the failure of the task of timely recovery, timely detection of problems, and provide information to the day-to-day maintenance of the cluster.

Ucloud Large Data Product system design

5. Future applications of large data in traditional industries

1 Intelligent Video Surveillance

Today, intelligent video surveillance has been applied to many places. For example, the Olympic Park in the intelligent counting and density analysis system, can be all-weather to the entrance area, the flow of people and other important areas of the flow of people in real-time density analysis statistics.

Intelligent video Surveillance, the core is the acquisition of video data, by frame extraction, applied to the feature model to match. These features are then stored, labeled, indexed, retrieved when needed, or analyzed more deeply. In this process, we need the original video data can be quickly converted into a feature model, but also need to mass storage to save this part of the video data. The core of the technology is the selection of feature model, the algorithm of transformation, the real-time and off-line analysis ability of massive data.

The selection of feature model can be established by communication with front-line workers, and some model libraries can be sought. In many developer communities, there is a sharing of conversion algorithms. Ucloud provides funding and resource support to some developer communities to help develop community sedimentation experiences to help more people. The real-time and off-line analysis capabilities of massive data can be built in-house, and can also consider the use of professional large data services, such as Ucloud's UDDP. Such intelligent video surveillance applications will easily fall to the ground.

2 Medical and health system

The future medical health system may be used for reference to the pattern of ID collection, taking photos uploaded to the designated website and then handling personal medical documents. Whether to go to a large hospital or to a community hospital for a medical examination, the data can be transmitted to the medical system. Subsequent medical treatment, the doctor can be based on the early examination data directly prescribe drugs, or based on these data for the diagnosis of patients. This can greatly improve the efficiency of doctors, to help solve the problem of medical difficulties. In addition, these data are gathered together and can be analyzed by the health and epidemic Prevention department to provide a better public service. can also be entrusted by individuals to commercial organizations to track the personal physiological situation, to do in advance prevention, aftercare.

Of course, the health care industry has a certain particularity, the need for institutional reform support. It is necessary to solve the isolated island of the existing hospital personal medical information in the informatization, and then extend the personal and public medical and health service with large data.

The change of the Times is irreversible from the beginning. When the internet was born, someone said, "on the internet, no one knows you're a dog." However, in the large data age, we not only know that you are a dog, but also know that you are always shy, front legs short corgi, do not like to eat dog food love to eat meat, the most like the pink small coat ...