In the big data era, Oracle helps enterprises move towards precise management to improve business value

In the big data era, Oracle helps enterprises move towards precise management to improve business value

Wu Chengyang, vice president of Oracle and General Manager of the Greater China technology product department, talked about how enterprises can use big data to help them achieve precise management. Enterprises need to integrate structured and unstructured data, in this way, we can obtain complete and accurate data to realize the business value of an enterprise. In addition to analyzing and predicting data, Oracle enterprise-level user big data can also protect users' investment.

What changes are big data bringing to the business world and people's lives?

For a simple example, we all talk about Precision Marketing, that is, Precision Marketing. For an enterprise, it hopes to find customers, what customers need is what they need. This is very important. Therefore, it is not only necessary to understand your customers, but also your purchase history in this enterprise, at the same time, it hopes to understand the customer's Social Behavior on social media. In this way, we can combine the two to recommend your favorite products. One of these aspects cannot be achieved, and many aspects are required. In this case, you have made a 360-degree attempt for this customer. Therefore, in Precision Marketing, big data is required.

Of course, big data also uses a variety of aspects, such as weather forecasts, such as smart transportation, all need some big data. That is why Big Data is such a hot topic today. In fact, the concept of big data cannot be concerned with big data as ordinary people do. People can concern about the benefits of big data. For enterprises, how can it help your enterprises. What's the ratio between your input and output? This is a key question. How much time do you need and how it can bring value to your company. This is not very important in relational databases. Why? After a relational database is established, your enterprise applications will naturally run, so your value is obvious. But this is not the case with big data. For example, you need to know that the data volume is large, TB or TB. Now you can see that, like Oracle, it can be 3000 tb in a cabinet, that is, 3 PB, but it can be done in a cabinet. If you say it's big, it's actually not big.

Therefore, I think the concept of big data should be called full data for enterprises. What is full data? Its volume is not the most fundamental problem, as a result, it has a combination of structured and unstructured data. You just can't look at structured data. You just can't look at unstructured data. You need to combine the two to make value to you. Just like the example of Precision Marketing for customers we just talked about, the problem you can see is that you need to know a lot about Precision Marketing for customers, such as banks, you need to understand all the records of this customer in the Bank. These are structured data, but you also need to understand the behavior of this customer in some society, after you understand these behaviors, you can decide whether I can give him a credit limit today. If there is no guarantee, I can double it. This depends on the analysis of all the data, so you can see that full data is very important.

Is big data reachable? How can we get value from Big Data?

Oracle has a good short film about fishing in the ocean. In fact, there are various creatures in the ocean. You may have tuna. You can see squid, seahorse, shells, seaweed, and various marine creatures. When you fish, after you capture it, there will be a variety of things here, but the value of different marine creatures is different. Its commercial value may be the highest commercial value of tuna, and I can sell the most money. I want all the fish caught by a net to be tuna, so I need to classify it. After extracting some of the most valuable things I have chosen and extracting them, you also need to decide, why, and which kind of marine creature on your market today, these foods and fish in the ocean are the most valuable in the entire market today, or the most valuable in future futures. You also need to decide which marine creature I want to choose based on the market changes, so that I can get the best price.

This is determined by the fact that I want to classify and extract the most valuable items. In addition, I also have a future price change based on the future, for example, your goods, based on future market changes, it is the best choice to determine which kind of matching I choose. In this case, we need to integrate structured data and unstructured data, and finally help your leaders make a judgment to help industry or corporate decision makers make a judgment, so that the value of the entire enterprise can be maximized.

So you can see that the most difficult problem in this article is how to integrate it. Today many people say that I can solve all the problems with Hadoop technology, for example, you can solve the same problem by using Hadoop technology to solve unstructured data. However, the problem is that you can use Hadoop to solve structured data. You just mentioned it. But in fact, if you must solve it like this, you will face a problem, you need to get the fire. You just need to make a little gas at home. But if you use Hadoop to solve the problem, you may need to drill the wood to get the fire.

So we think that today you should use Hadoop to solve unstructured things. The important thing is to load the things you are interested in into structured data through the loading method. Then, analyze and judge the structured data and obtain the expected results. This is a very important point. In fact, its difficulties, fundamental difficulties, and most important difficulties are its loading and integration. It is not the technology of Hadoop itself, because today, from the perspective of technology, Hadoop is a relatively mature technology.

In fact, another question is that I understand that big data should be a structured and unstructured fusion today. The difficulty is that I should load it up. I encountered another problem: Can my own people come to my own unstructured data? In this regard, the challenge is that the traditional architecture for solving Hadoop today uses a technology called MapReduce, relatively speaking, there are many people in the Internet industry, but you use less for enterprise level. The question is, should you hire so many people to solve MapReduce, learn this technology, and then solve the problem? This is a huge investment for you, so today you can find a new technology in the world, we use SQL to extract data from the Hadoop architecture. This is very interesting. I use a method that you are most familiar, in one of the most traditional storage methods of unstructured data, to extract things you are interested in, you do not need to learn MapReduce, you have no need, you only need to understand SQL.

Everyone said that big data is the most important thing to be comprehensive. This is indeed the case. In fact, you will find that today, for example, if you require hundreds of TB and hundreds of TB, we require real-time results. That is to say, if you do not reach a p-level, TB structured data, you should require real-time, that is to say, if your leadership says today that I want to judge whether I am a fishery company today, if I increase the production of tuna more than doubled today, I will focus on the entire market in the future, how much has changed my overall revenue this quarter. He can get a structure immediately without waiting. The important thing is that you need real-time, you need very real-time, and all of you need to be in the structure, of course, you can use the so-called memory technology. You will find that many technologies are actually integrated. Today, big data is not a technology that completely denies other technologies, it is a big data that works well with other traditional technologies. This is probably our understanding of big data.

How Does Oracle help enterprises get value from the ocean of big data?

The entire way of thinking for enterprise-level user big data is different from that for the Internet. Why? Because enterprise-level your entire data value is 70%, 80%, or even 90% structured. We once discussed with a bank's CIO that the bank's CIO told us why I used Hadoop technology? I have cleaned structured data many times. I have already done a great job and can solve it with traditional technologies. Of course, we also need to explain that you still need to find some unstructured data on social media in many places. To some extent, 80% and 90% of the enterprise's greatest wealth are structured. But today you need to integrate unstructured data from the market, from your social media to the Internet, and so on. Oracle is aiming at such a development trend. As you can see from Oracle, we use Hadoop Technology in the unstructured aspect. We use Hadoop of Cloudera, it is the current commercialized Hadoop architecture. At the same time, if you use MapReduce, we fully support it. After completing this process, we have an Oracle Loader. Oracle Loader can extract unstructured data. When you extract what you call the essence, load it into the structure. The Loader is faster than five times the traditional speed.

Loading is critical. If you load slowly, it will be useless even if it is earlier. You said that I still don't understand MapReduce. As I said just now, Oracle invented a Big Data SQL statement. It doesn't matter if you don't understand MapReduce. After installing it based on Hadoop, you can use Big Data SQL to read the Data and put it directly in the structure. In the structure, you have a lot of imagination. You can perform data analysis, data mining, data prediction, and data warehousing, all these traditional tools can be used, and your investment in the past is protected. Therefore, we think this is the best solution.