The cold thinking behind the big data surging

Big data surging, popular IT industry

It industry has never lacked a new concept, term, new technology, update fast, the launch of the fierce, dizzying.

After cloud computing and social networking, big data is now on fire again, surging, and seems to be the most fashionable word in the 2012 Information technology field.

Oracle, SAP, IBM, Microsoft and other it trolls, like looking for a new gold mine, began to dig up large data, multi-directional promotion of large data concepts, especially SAP's Hana and Oracle's exalytics is hard, fighting for "head hot soup." And a large number of small and medium-sized It manufacturers also followed in droves to share a large data market a cup of soup.

What is the big data? According to the IDC definition, large data means a new generation of architectures and technologies that are designed to be more efficient in acquiring value from high frequencies, large capacity, different structures and types of data, are used to describe and define the vast amounts of data generated in the age of information explosion, and to name related technological developments and innovations.

And compared to the sudden rise of large data, the former business intelligence analysis is well-known BI developers is difficult to cover the embarrassment and loss of the situation, and even forced to retreat to the corner. In recent years, large data has brought a great impact on BI (Business Intelligence Analysis System), and the pace of development has been greatly dragged down. It has been boldly predicted that the next decade, the business intelligence analysis of large data will lead the development of management information.

From a variety of professional reports, analysis, we can see that large data presents 3 kinds of characteristics: Volume (data volume), velocity (processing speed), produced (data type). Volume refers to the large amount of data, now many enterprises have been faced with a single day data volume of dozens of, hundreds of TB (trillion bytes, 1TB=1024GB) of the increase, and the total amount of data reached the PB (Petabyte) level, such data volume has made the traditional database difficult to deal with; Velocity refers to the increasing speed of enterprise data, a wide range of applications, such as mobility and social networking, have increased the speed of data growth faster than traditional enterprise applications, and the faster the data is proliferating, the quicker it will be to keep up with the data processing and analysis, while produced refers to the diversity Nowadays the internet is not just about reading information, at the same time also constantly in the output data: Upload photos, upload video, hair microblogging, on the other hand, it all over the work of life in all corners, a variety of sensors, monitors also constantly produce a variety of machine information, the data type has become increasingly complex and diverse. This has spawned a strong demand for large data technologies.

From a number of mainstream manufacturers of the product introduction, we can find that the main difference between large data and bi is that, compared with the traditional transaction based Data Warehouse system, it can carry out more data and non organization data processing on the basis of BI, and the large data analysis not only focuses on the structured historical data, but also prefers to the web, Social networks, RFID sensors and other unstructured mass data for better analysis, overall compared to BI, large data is a perfect big upgrade. Internet companies such as Facebook and Twitter, which face big data explosions, have begun using new technologies such as Hadoop and NoSQL to solve massive information problems and have achieved some success.

Big data, bi present?

Therefore, how to solve the increasingly urgent large data processing has become the inevitable demand for enterprise management information and modernization. But how much activity is there in the country's big data field? is the big data really as powerful and useful as some of the manufacturers are going to be?

In a cheering sound, some experts and insiders appear cautious, even there is no lack of criticism. Some experts believe that, in addition to a large number of seminars, and various companies to declare their ambitions to enter large data areas, the actual progress has been difficult to see results. Many CIOs believe that the industry's value behind the big data is concentrated in finance, telecommunications, energy, securities, tobacco and other super large, monopolistic enterprises, other industries to talk about the value of large data is too early, the use of large data in enterprises is not that as long as the Open data, the use of some technology can easily find "gold ”。 At present, large data applications in China seem to be showing such a state: Investors are active, technology and service providers enthusiastic, digital media high-profile, and a large number of application enterprises confused.

Some experts believe that, from the results, the question of large data is no more than Bi, the same encounter the "one kilometer short" embarrassment. It was said that "big data is a relative concept, packaged in existing schemes, and is treated in new bottles of old wine, but more fashionable." "Data applications in the mass data age have not revolutionized the number of enterprise digital operations, and before MapReduce, Hadoop (both a programming model for parallel operations on large datasets), companies can easily scale parallel computations of data, while NoSQL The advent of the data is only the way to bring more possibilities, there is no revolutionary qualitative leap.

From some companies in the industry to take out the big data application examples, still only in the traditional sense of data analysis and BI, but cleverly put the account on the big data. One developer said that using its big data technology, an E-commerce website could know "where people buy things the craziest" or "what type of phone is best to sell", which is the result of big data analysis. The expert retorted, "is the same result of the BI analysis based on the Data warehouse system different from the results of this large data?" It is true that new values are mined from some large data applications, but this value is only added value, there is no reason to exaggerate it, and there is no reason to imagine. Big data is opportunity, but only a few opportunities, more of the giants of the business strategy. "A CIO at an electronics company in Xiamen also believes that" the data content and usage that some companies need are actually available through the open source community, and traditional column data can handle large data well. To participate in all kinds of big guys in the mouth foam flying meetings, and engineers to talk about what can be used more practical tools to specific practice, play. ”

For example, EMC, which promotes big data, does not have much of a change in its Greenplum core product line-still divided into Greenplum Database (Data Warehouse), Greenplum HD (Hadoop analysis) and Greenplum DCA (Data computing equipment), the latter is based on cost-effective industrial standard x86 server MPP (large-scale parallel processing) distributed extensible architecture. Therefore, standing in the manufacturer's perspective, if there is no more new and meaningful things, too much investment in resources to promote a large number of publicity is obviously not cost-effective to prevent the final customer does not buy. Indeed, there is value in the unstructured, semi-structured data of massive growth that deserves deeper digging, but that does not mean that people will have to replace new methods and tools to deal with them at once. Just as demand is incremental growth, business change should be more gradual and prudent.

What eye-catching hang what label, what has the advantage to which drill, this is now the tricks of the merchants. In the face of the "big Data" of the popular, many traditional bi manufacturers can not help but finally "temptation", have shook his head a change, are set on the "Big Data" jacket, it is exclamation.

It can be said that the current it manufacturers mentality increasingly impetuous, quick success, not really to seriously study customer demand, seriously study business management, but still stay in the hype concept even cashing to go stage. Today you an SOA, tomorrow I a EAI, today you come to Grid computing, I will come to a cloud tomorrow, you a bi, I come to a large data, and so on, are their respective technical characteristics to explain the concept and application of their software, guide users to throw arms, But not a few can say that their software services in the end is a good thing, what can give enterprises a simple and practical benefits? is the best price/performance ratio? Instead of a variety of concepts, definitions, customer units dazzled, overwhelmed. In the choice of time to see this also a bit of truth, the talk is good as right, but no manufacturer of the system really make customers deeply satisfied.

Come back and say big data and bi. It can be said that there is a deep natural connection between the big data and the BI, the twins, the office decision-making itself is a kind of teamwork and coordination, especially in the data mining and data analysis level, and not much difference. At the same time, the relationship between traditional bi and large data is not substitute for each other, exclusion of the relationship, they are like the left brain and the right brain, the division of labor, the traditional bi to deal with structured information, large data to deal with unstructured, semi-structured information, they are interdependent, complementary, altogether as a whole, the formation of a complete enterprise information brain.

The innovation of large data, advanced and forward-looking, can not be denied, it is worth affirming, but when someone proposed "big data, bi present" theory, it seems too arbitrary, extreme. In today's segmentation of the winning era, the function is not the more the better, the function is too much to be redundant, increase the deadweight cost, thus the myth, knowledgeable the concept of excessive speculation, but has lost the essence, primary and secondary. "Big data will make bi more valuable and business-friendly," said Ritasallam, BI analyst at Gartner Research. We always need to look at past data, and when you have big data, you should do it. BI does not disappear, it is strengthened by large data. In a certain period of time, large data is difficult to replace the traditional BI tools. ”

Today, there is a constant confrontation between big data and BI software who are superior to each other, however, it should be understood that customer units, consumers really need is not the concept, the need is not pros and cons of the controversy, they need to be real application software, need to solve the problem of effective methods, What is needed is just the right functionality of the software.

But for the application enterprise, they must weigh carefully, in the end enterprise utilizes the big data to be able to bring to the enterprise how many added value? Can this added value make the enterprise's investment a better harvest? And more importantly, whether the use of large data will be able to give the enterprise can not be achieved before the value of ? All these need to be considered carefully by the application enterprise.

Http://blog.sciencenet.cn/blog-549158-753166.html