Big Data: A new opportunity or a piece of paper?

Now, some people think that the value of big data is to help businesses get new insights and action, while others think Big data is just hype. The big numbers are opportunities or empty words, what is the mysterious boundary between the two views?

At present, people have different cognition about large data and its value. Some people think that the value of large data is to help businesses gain new insights and action; others think Big data is just hype. Both of these ideas have merit, and interestingly, both are true. Despite the hype surrounding big data, people quickly learned the difference between the real value of big data and the rhetoric.

Figuring out this distinction would be very helpful in understanding large data values (preferably considering investing in large data) and recognizing the challenges that still pose major impediments to most enterprise development. Let's assume that future technologies will gradually mature and create value by releasing their potential. This prediction has been proven in many previous technologies, and large data technologies should be no exception. The main bottleneck constraining the development of large data technology is its own problem: people will ignore the serious dependence of large data, or think that dependence is only the prerequisite we must accept to create value.

This dependency refers to the need to maintain data consistency before creating value, or to standardize data in systems such as databases, which require companies to invest billions of of billions of dollars, which leads to inefficiencies and duplication of effort. Therefore, before creating any value, the enterprise project investment has reached 70%, for data recognition, acquisition, migration, storage and optimization. Although analytical technology has made significant breakthroughs over the past decade, and the number of analytical terminals and platforms has soared, the development and deployment patterns of enterprise analytics solutions have not changed over the past 30 years.

It is a matter of concern that in industries where large data technology is truly creating value, the large data market is showing a trend of continuous segmentation, and this trend is becoming clearer. We'd better start with these market segments and get a deeper understanding of the difference between big data values and empty talk.

Areas that can embody large data values

Large data technology has been successfully applied in the fields of data exploration, trend analysis and adjustment opportunity analysis. This seems unquestionable, and the following two common points are not obvious, but large data technologies are already viable and firm in areas that are in line with these commonalities.

• A new mass of interactive information: web-based shopping and digital retailing, mobile-side activity, social media interactive information and Internet search terms. In other words, a new mass of similar data.

• Attach importance to Marketing opportunities: for product sales to enhance the potential customer identification success rate, this technology application is usually by popular marketing and media costs.

A field that fails to embody large data values

As the data of the same kind is reduced, the cost of gaining insight increases correspondingly, the value of large data begins to decrease, and the hype of the comprehensive value factor of large data is led astray. When it comes to typical business issues, there are few successful cases of big data. Why?

• Business issues are commonplace. This is no longer necessary, and the use of "new" data over the past 5-10 years is one of the success factors in the successful deployment of large data technologies.

• Solutions use different types of components. The challenge for enterprise data is that it is widely distributed across a wide range of technologies and data platforms. For example, digital retailing, telecoms and social media use structured data in a similar manner, while enterprise data is distributed across host, ETL (extraction, transformation, and load) tools, virtual tiers, relational databases, business intelligence (BI) databases, transaction databases, and hundreds of other components. These technologies have been evolving over the past 30 years. Worse, with each application using a different data model, resulting in more complex integration of data with its associated technology platform, it is difficult to create direct value using the current large data tools to access enterprise data.

This is why most business problems in an enterprise are not related to large data. These business problems are actually distributed data problems: In this model, information, data, value and analysis are widely distributed in different locations, technology platforms and data sources. However, we continue to use the same centralized model as before to solve this increasingly serious distributed problem. These centralized models can play a big role when users are able to access data stably through the common appearance of the interface, as is often the case in new success stories in industries such as social media and digital retailing. But the centralized model does not solve the banking, insurance, healthcare and other broad business problems.

Standish Mellon Asset Management company recently updated large data project successful case research report. The update shows that only 6.4% of the 3,555 large data projects with a labor cost of at least $10 million in 2003 to 20,120 years were successfully completed. This is the result of an enterprise's inability to efficiently manage distributed data sources and their associated technologies. We still need to use large data technology to overcome these difficulties.

At present, the enterprise realizes the large data value needs to carry on the data integration and the standardization plan to many different data and the function system. Without changing existing data management mechanisms, the more distributed components The Enterprise solution employs, the lower the project returns.

Large data technology to promote in-depth analysis and analysis of performance to obtain technical breakthroughs, its value is beyond doubt. But this value is corrupted by data extraction and/or integration costs, leading to the bottom line of value/hype being easily breached. At present, there are some differences in data value of the market, some of which are still in the start-up period, can maintain technical consistency, so these industries can temporarily solve the problem of distributed data.

Because the technology islands will continue to exist and the data remains in different locations, Gartner analyst Doug Laney predicts that by 2017, 90% of the big data items will still be unable to play their part. Doug recently concluded that while data complexity, the distribution and dispersion of internal and external data are rising, there is no recognition of the rationality of systems and data large-scale integration projects due to the various commitments made by large data technologies, which are only the embodiment of large data values.

We only realize that although large data technology has its own place, it is still affected by the distributed data source, so we can make full use of the data value according to the data complexity and distribution.

Most enterprises have the flexibility to use dual data strategies: Use large data technology to conduct in-depth analysis and opportunity identification of a large number of similar data, or to use distributed data to address complex but already understood challenges to operations, risk, and management. People will be able to accept this dual data management strategy, fully identify, explore and manage the value of large data technology, and in the industry to achieve constant flexibility in innovation.

(Responsible editor: Mengyishan)