Ten important facts about big data

Whether you favor or reject it, big data has become a fact. Now we have to go back to the facts and explore the truth ...

Big Data is one of the hottest topics of the day, and none of us can be in it. Like the cloud that emerged a few years ago, big data has caused widespread concern in the marketplace, and there is an urgent need for companies to define big data. Large data lacks a standard and pervasive definition, at least not as widely accepted as NIST's definition of cloud.

The definition of IDC, a research firm, may be more easily accepted. Its definition of large data is: a new generation of technology and architecture, with efficient capture, discovery and analysis capabilities, able to economically from the type of complex, large numbers of data mining outstanding value.

Big data has become an important issue in various conferences, and executives are reluctant to miss out on this emerging trend. There is no doubt that large data technologies will be used when future companies try to analyze the vast amount of information available to drive business value added.

On the other hand, as with other emerging trends, many people doubt the usefulness of large data. In fact, when a technology becomes the focus of widespread controversy, some doubts and criticisms will surely be incurred.

There are two distinct points of view about the important value of large data. The common point, however, is that both views have some misconceptions about big data and obscure the nature of large data.

Misunderstanding

Myth 1: Big data only means huge numbers

The name "Big data" is inherently misleading, as if the size of the database is the problem. But that is not the only factor. Alan Priestley, director of strategic marketing at Intel Europe, the Middle East and Africa (EMEA), argues that there are other elements of big data, most notably the complexity of data types and the need for fast delivery of data. In addition, the enterprise also needs the first time to understand the accuracy of the data.

Myth 2: Social media is the most important

Much of the discussion about big data is focused on the impact of social media data on businesses. It is not hard to understand that the majority of media focus on the traditional business of getting the latest information from customers. Now, it means looking for social media interactions, Twitter, Facebook, Insta-gram, and so on. However, Priestley pointed out that the most common enterprise is the machine-generated data, including blog, data center log and other information.

"The aviation industry can now also use the power of big data," he says. For example, they can use and analyze air travel
Data to predict possible problems. In the past, they had to check the engine only after hours of flight or failure. No one wants the fault to happen, but it's too late to check the engine until after the crash. With large data analysis, they can track the vibration of the engine. By examining the generated data, they are able to issue alarms and schedule inspection engines when the data is found to be abnormal. ”

As an example, Priestley also describes how BMW successfully exploits large data. BMW's large number of cars can access the Internet through 3G technology. By using large data and related analysis capabilities, BMW can track these cars and contact the owners. Of course, there are a number of examples, such as credit card companies that can check fraud transactions in real time, ensure that remote purchase transactions are legal, and all of these operations take only a few seconds. Intel itself is also an important user of large data technology. The company uses large data to control the benefits of wafer manufacturing plants, dramatically reducing costs and reducing waste.

Myth 3: Big Data is Hadoop

Many of the big data discussions are focused on Hadoop. The Apache project is of course the highest profile, and it is the first tool to analyze and store unstructured data to gain value from it. However, it is not the only tool. Priestley said: "Some people think that as long as the use of Hadoop is nothing to worry about, in fact, the traditional data warehouse still have space." People need to keep their existing IT infrastructure. ”

Priestley points out that the attraction of Hadoop is that businesses can get a lot of information with only a small overhead. He added: "You can download Hadoop in Apache, a freeware software that can be run on a standard server." Another alternative is to buy integrated solutions for companies such as Oracle or Teradata. But for many companies, this may not be a viable option unless they are fully aware of the advantages that can be gained by analyzing the data. ”

Myth 4: Want to quantify the return on investment (ROI)

Companies like hard numbers. CIOs generally like to say that the cost of migrating to large data is X, which will save y in three years. In fact, the big numbers are not. It is very difficult to get a clear return on investment (ROI) from a large data plan. As Priestley points out, a large number of large data implementations are "hypothetical information" that is difficult to define.

The impact of customer relationship management (CRM) on the enterprise can be quickly measured. But unlike this, companies planning to adopt large data must accept the difference. In addition, the way companies think about investment returns (ROI) on major projects seems to be changing. In the past, businesses believed that ROI was always a tangible asset that could be easily measured, and that business advantages were bound to exceed the cost of spending. But things are starting to change.

Recently, claranet A survey of cloud migration in the enterprise. The results showed that more than One-fourth of respondents viewed ROI as one of the decision factors, while 79% per cent thought that ROI calculations did not reflect business advantages. Although the survey focuses on cloud migration, it is reasonable to speculate that large data migrations will not be much different. Both represent a technological leap forward in the future.

Myth 5: Results are not guaranteed

Big Data is an unknown. What you are doing is analyzing the numbers that are incalculable and difficult to determine. In essence, large data is not easy to understand or abstract. Otherwise, you won't need large data technology. Therefore, enterprises must realize that they cannot guarantee the accuracy of the results. It is futile for companies to try to obtain results and to find hypothetical supporting data. In the example above, the airline may want the aircraft to be maintained once every 500,000 flight hours, but if the plane crashes every 200,000 hours, the airline's vision will be meaningless.

If there are some misconceptions about big data, some of the key facts about big data need to be seriously understood by companies that are not too bullish on big data.

Key facts

Key fact 1: Need different skills
Most observers agree with the shortage of data scientists. By 2019, McKinsey predicts, the world will be short of up to 190,000 scientists with large data. The reason is not difficult to find. Processing large data items requires a completely different process from the existing Data warehouse
Skills. And it is not limited to data processing, but also requires the ability to convert data into actionable recommendations.

"Hadoop has a tool called Map Reduce. It requires Java programming skills, which are not the skills of many data analysts today. Priestley for example. And it's more than that. The ideal person to work with large data also needs to understand business processes, Java, and statistics, and may even require some SQL skills. This is a big problem, so many people agree that the shortage of data scientists will be an important impediment to the adoption of large data technologies.

Key fact 2: Identify your goals

While companies should not try to explore the results, they should be clear about corporate goals and a goal to be achieved. Cases
For example, one way that large data can improve performance is to gather more accurate information, including personal data, customer behavior, and purchase decisions
Policy.

The McKinsey company found that the numbers were staggering. The business consultancy claims that if the U.S. medical industry uses big data
, America's health care costs will be cut by 8%. In addition, McKinsey mentions that by reducing fraud lawsuits and increasing taxes, European public
The department can save 100 billion euros in operational efficiency.

Key Fact 3: Man is the driving factor

Big data projects need someone to push. Technology is not a critical issue. This does not refer to those who have the skills of the data scientists mentioned above, but to those who can put forward clear goals and needs and can implement decisions.

These people do not need special management skills. These responsibilities may fall on the shoulders of the chief Financial Officer (CFO), Chief Information (CIO), or even the chief executive (CEO), but in the end, one person needs to take this responsibility. As Priestley points out: "Big data is not just a technical challenge, it's a business challenge." Businesses need to understand this. In this respect, the usage pattern is very important. In this regard, enterprises can have a variety of patterns, and in different ways to model. ”

Key Fact 4: not just data

Large data analysis has three main elements: the data itself, the data analysis, and the presentation of the results. Owning the data itself is of no practical significance. The data itself already exists. It is important to process, analyze, and present important information to transform data into important values. The development of large data projects requires careful planning. It's best to start small, implement a single project, and then scale it up gradually. Detailed results analysis is required after data acquisition.

Key fact 5: Large data covers all

Much of the discussion of big data is focused on large organizations, and the stifling mass of data for these large bureaucracies has constrained the effective functioning of the Organization. Many of the first organizations to adopt large data technologies fall into this category, but they are not the only beneficiaries.

All kinds of businesses want to get skills to evaluate hidden data and induce patterns. Some small businesses need to deal with large amounts of industrial data. Formulaone design, for example, is small but manages a huge amount of data, so even small businesses can benefit by using large amounts of data in their daily work.

These companies may want to go beyond Excel to analyze customers and look for customer buying patterns. For example, if you had a special fish on your restaurant menu, it was canceled. So when this dish appears on the menu again for the customer to use, you can use email to notify all customers who have ordered the dish before. Or, if you are a wine merchant, there is a wine in your inventory that is brewing, and when they are about to come out of the library, you can remind the wine lovers.