Big data is a "double-edged sword" with a thorn.

Objective

In the 2012 "Big Data" is gradually appearing in our vision, to 2013 years "big Data" has become the hottest topic of discussion, then what is the big data, the big data in the end have what magic to hot? Let's look at the definition of large data: "Big Data", Or the vast amount of data, refers to the volume of data involved in the large scale to be unable to pass the current mainstream software tools, within a reasonable time to achieve capture, management, processing, and collation to help the business decision-making more positive purpose information. Large data in the large data age, prepared by Victor Maire-Schoenberg and Kenneth Couqueil, is a 4V feature of large data, using a short cut of random analysis (sample sampling): Volume (Mass), Velocity (high Speed), produced (multiple) , veracity (authenticity). This paragraph about Big data source Baidu encyclopedia.

In fact, "big data" is not 2012 years before the birth, as early as the 90 's when the Amazon, Google, Microsoft and other companies have begun to dig data, Now we can see that companies that are now using and using large data have been able to lay out and study data for many years before, for example, IBM is now the world's largest information technology and business Solution Company. As early as 2011, IBM launched infosphere large data analysis platform, such as Google, BigQuery is a Google launched a Web service to handle large data in the cloud. The service allows developers to use Google's architecture to run SQL statements to operate on super large databases. BigQuery allows users to upload their oversized data and interactively analyze it directly, eliminating the need to invest in building their own data centers.

This shows that "large data" in fact, relatively early in foreign countries have begun to rise, only to the domestic time is relatively short, with "Big data" in the domestic rise, some SMB enterprises also began to pursue large data, want to be in the trend of large data to get their own that bucket of gold, but for the small and medium-sized enterprises seemingly beautiful "big data" Success still has a certain threshold, rational view and scientific use of large data applications for SMB enterprises is the right way.

Note: SMB (SGT and Midium-sized Business) refers to small and medium sized enterprises with limited operation scale, personnel and funds. Countries have different definitions of small and medium-sized enterprises, some are divided into the number of employees, and some are divided by turnover or market share.

2 Big data on the lure of business

For enterprises obsessed with "big data" there is always a reason, especially in some industry experts under the bluff is to let some small and medium-sized enterprise owners think large data is very magical, so it is crazy to pursue the application and use of large data.

The temptation of large data

So the big data for the enterprise in the end what are the benefits, here for everyone simple summary:

First, large data mining can help enterprises understand customer needs

Can large data be able to analyze the attributes of an enterprise user base to enhance the enterprise's demand for users? The answer is "yes", which is the charm of big data. Large data sources for enterprise employment but these data seem to be dispersed and unrelated, but reasonable and valid reasons data model for modeling analysis you will find a certain user behavior rules and trajectory, through the user's these habits, enterprises can be based on their own business to push the relevant services to users.

Here is an example of how the enterprise uses large data to understand the needs of users. For example, your client likes to ride outdoors, will be installed on their bicycles a number of monitoring equipment, through these instruments can detect a number of cycling conditions, through such data accumulation, enterprises can be in the user's cycling around some of the traffic conditions, repair shop information or store content, user-friendly, And through the user's behavior to judge the nature of the user, in the appropriate time to push the appropriate content, frankly speaking is to allow enterprises to do precision marketing, of course, this is in the mobile end to reflect the value of large data.

Second, the application of large data can save enterprise time

For today's enterprises, regardless of size to save time is to save money, but in some huge amount of data to improve their efficiency this has been a problem, because how to quickly deal with these data they do not have a good solution, but the application of large data can effectively improve the enterprise and reduce the time of the enterprise. Wal-Mart, for example, has designed the latest search engine Polaris for its website, using semantic data for text analysis, machine learning, and synonym mining. According to Wal-Mart, the use of semantic search technology has increased the completion rate of online shopping by 10% to 15%. "For Wal-Mart, that means billions of dollars."

Third, the big data is helpful to the enterprise development

For the enterprise, the big data meaning is not to grasp the huge data information, but is the specialized processing to these meaningful data. And at present, many small and medium-sized enterprises in the country they have a large part of the data is still in hibernation or half dormant state, and did not produce great value. For businesses, large data also helps streamline business processes, for example, by leveraging social media data, web searches, and weather forecasts to unearth valuable data, the most widely used of which is the optimization of supply chains and distribution routes. In both areas, geographic positioning and radio frequency identification tracks goods and delivery vans, using real-time traffic route data to develop more streamlined routes. The human resource business is also improved through the analysis of large data, including the optimization of talent recruitment.

From these simple examples we don't look bad. In fact, "big data" for enterprises is still very useful, whether in precision marketing or enterprise internal process management has very important significance, but for some small and medium-sized enterprises to pursue large data mining will be so easy and simple? Actually, For some enterprises for "big data" potential is to have to pay.

3 Difficult one: traditional enterprise IT architecture can not adapt

As we all know, the big data is simply the data that the enterprise stores, and the data stored in the computer is counted by byte, so how many bytes should the large data be? We simply looked up some of the current enterprise data storage, we can understand.

Enterprise IT architecture

It is reported that the current Internet enterprise data volume reached 1000PB; in the energy industry, only China's national power grid smart meter data up to dozens of PB; in the medical sector, the health records of a large city are up to 5PB a year; Meteorological satellites and weather radars can form terabytes of observational data daily. According to statistics, 2013 China produced more than 0.8ZB of data, is twice times 2012. From these data we are not hard to find, in fact, we have now entered the era of information and data explosion, each industry has multiplied the data.

It is precisely because of the exponential growth of data for the enterprise storage has become a burden, simple storage equipment can not meet the needs of the enterprise data storage, coupled with the huge amount of data to search, backup and other applications have brought great burdens, the traditional IT architecture has been unable to meet the needs of the times. We take the products provided by SAS as an example, their product price of 100,000 U.S. dollars, for the average enterprise 5-10 can meet the demand, to five as the basis for the enterprise equipment basic expenditure nearly 7 million yuan, this also does not include follow-up upgrades and maintenance costs. Therefore, the seemingly beautiful large data applications in fact for the limited funds of the enterprise is definitely a good vision, once put into it, the use of improper is absolutely a bottomless fund.

4 Difficulty Two: Data Model complex data realizable value is very difficult

In fact, for enterprises, pure data will not produce value, how to use these data to be processed and put into practice will produce value, in order to make the data visualization, so it is necessary to use the data model for data modeling and analysis, so the data module is very important for large data.

The model of large data is complex

Different enterprises have different data, such as personnel data, Web text, transaction data, call data, sensor data, mass audio video, and the current structured data in this data accounted for only 5%, semi-structured data accounted for 10%, unstructured data accounted for 85%, So for the enterprise how to let these complex data types for integration, management, analysis, to achieve the maximization of data value becomes a big problem.

At present, the mature and effective data model is limited, only to meet the application of some enterprises or industries, so for many small and medium-sized enterprises if there is no appropriate data model for their data analysis and collation to make use of large data is also very difficult. And the data models have their own advantages and disadvantages, and they apply to different fields. Regardless of the model, the choice is based on the actual application scenario. In particular, for some enterprises, a single data model can not meet their own solutions, many large applications may need to integrate a variety of data models.

5 difficult three: data independence strong connectivity poor

When we analyze the second difficulty, we know that there are many kinds of data and the structure of the data is different, and there are some differences in the data sharing process due to different database, such as different operating system, database type, hardware platform and so on, to form some data independence, This has been a hindrance to the sharing and running of data. So how to realize data integration and sharing among heterogeneous databases becomes an urgent problem to be solved.

Data isolation

For enterprises, "Large data analysis" is to rely on data analysis to find out the problem, and through the model and forecast analysis and improvement to develop, to achieve enterprise in the industry reform and innovation. At the same time, our blind reliance on data leads to rigid thinking and decision-making. When more and more things are quantified, people are more likely to fall into the myth of valuing data. So how to avoid being a slave to data is also a very important subject.

6 difficult four: How to balance data security and personal privacy

We said in the front that the effective application of "big data" can enable enterprises to achieve accurate marketing and improve the process of the enterprise or the future of the contract, but in the face of large data privacy issues are directly on the table, especially at present individual users of personal privacy gradually attention, which also let enterprises into a difficult to balance the situation.

Security and privacy are key

For enterprises, the effective use of e-commerce, search engines, SNS social applications and other Internet service providers can be user behavior data mining and analysis, so that can be achieved in the future precision marketing or to achieve some commercial interests, But through these means and the form of active user information is unavoidable to involve personal privacy. It is a basic guarantee for enterprises to make use of large data analysis to avoid the disclosure of user information and ensure the security of data.

Not only that, if you meet some illegal enterprises to use large data can also predict and control human potential behavior, in the absence of effective ethical mechanism may cause some illegal things happen, so large data for some enterprises to apply also involves the moral bottom line problem.

7 Summary

For enterprises, the core value of large data is the storage and analysis of massive data, through the reasonable and effective analysis to carry on the related change and the preliminary judgment, compared with the existing other technology, the big data "the cheap, the rapid, the optimization" these three aspects comprehensive cost is the optimal. Especially in the tide of the internet to promote the "big data" in the future will certainly have better prospects for development.

For enterprises, large data may not meet the needs of all enterprises, but it does give enterprises the opportunity to develop and improve revenue, but also how to data collection, processing, analysis are the actual problems faced by enterprises.