Yianyang: The Evolution of enterprise data center and the application of securities industry data

The 2nd Session of the "2012 China Data Center Industry Development Conference" was held in Beijing on the 20,112-year 17th, in the afternoon of cloud computing and the flexible and efficient IT Infrastructure forum, Dr. Yianyang, general manager of the China National Securities Information Technology Department, made a keynote speech on "The evolution of enterprise-class data centers and the application of securities industry data". The following is the full text of the speech.

Dr. Yianyang, general manager of China National Securities Information Technology Department

Yianyang: Good afternoon, everyone. I come from the financial industry, I talk about cloud computing from the user's point of view, and then I put our efforts to promote cloud computing, especially the use of large data, especially last year, I presided over the securities industry common cloud related work for everyone to do a share.

Enterprise-class data centers, at the foundation level, many companies are working on cloud computing, including the process of virtualization. It's not difficult to do it in detail, but the difficulty is to make it more complete, and a lot of standardized and standardized work, such as our progress in the (English) process, due to cloud computing and the introduction of virtualization, (English) Some of the standard changes in the enterprise to promote cloud computing in the middle of the process, there is a work, In the past few years in our financial enterprises to do more, the first to establish an enterprise-class data center, because resource sharing it can put many functional departments, many business chain through the way of business integration, may bring some impact on the business. This is a summary of our work over the years, in the process of pushing is very difficult.

First, why set up a data center? It is the necessity of centralized management. Includes management, including business, including technology, which is a complete integration.

The second separation of the appeal, in our process of advancement, not only to have integration, but also in accordance with the rules of various aspects to be separated, otherwise the efficiency is not the most satisfactory.

Third, the unification of data makes the exchange between each other a basis.

Four, through data integration, we can get related business innovation, a lot of data is buried in our underground resources, must be unified mining, in order to make our implied value of the business can be dug out.

The basis of decision making is to provide relevant basis for decision-making department.

Second, how to build enterprise-class data center, first of all, this data center must be efficient, how to build an efficient platform? Must choose a mature platform, now a lot of technology are walking in the forefront of the business, the key is how we use the existing mature technology for our business services. The data model is also very important, many of the application system evolved into a data center, its problem is that the expansion of the difficult. The previous data centers were triggered by the public system. After all, it is the control of our existing user-related data operations, and the operation of our existing personnel, our potential customers must be included in our business, so we have to introduce the financial model of banks, the establishment of data centers is very important. Data centers must be a process of distribution and construction.

The third principle, we have the content of the data center, first, to collate data, the second with our data to provide services. Third, we can provide support for our business through data integration, and provide the basis for our decision. We're in the process of building your IT department to push a project, or to build a huge data center, it has to pass the project, do the data center is relatively abstract, and in the short term will not produce benefits, as we are doing a high-rise, the initial excavation of the foundation when you do not see its appearance.

You must have a very good reason for the early stage of the project to advance the work. So this 3W principle is a complete description of the middle of our data center building in the years ahead.

What kind of platform do we choose to build in an enterprise-class data center? What is our IB? What is our ETL? At that time 08 years of time we can choose very few things, now Kaiyuan is a lot of things, there is much room for choice. We are in China, in the financial industry, the first to introduce a NPP architecture such a data center, data warehouse Such a framework, now our entire operation efficiency, is the query time can be hundreds of millions of records in a few seconds can be out, and you are in the data slicing more complex, time growth is not much. So let's embark on a very bright road to docking with the future of cloud computing.

It is a complete solution, we are in the process of propulsion, the choice of technology platform it is very important. The information application of the big finance, the bank still goes to compare the front, we are the financial data model with the securities industry to do a fusion, this to our innovation business development, basically do not need to do any modification.

This is a customer experience, many of our tools are aimed at our technical personnel, we want to open to our customers, we must consider the customer experience, we have a number of tools and development platform, we can use our customers very good experience of this tool to get the relevant application of the display.

There is one more concern in the Enterprise data Center, is the security, the first is the number of our data management, it is likely that the data is incomplete, the data information error situation, so you take what mechanism to ensure that our data is complete, consistent, and accurate. In this design, there are related skills in it.

Second, the data from each node scheduling, it must have the corresponding rules. For example, from the data source to our core data layer, and then to our data warehouse, there are relevant rules to promote.

MPP architecture is also our current in the cloud more popular in the MPP architecture, including Sina's major web sites, the need for analysis of the architecture, are the use of such a pattern. We now use (English) to do this data warehouse, on the use of (English), we used two nodes, like Taobao has nearly 280 nodes, our two nodes basically for our neutral brokerage enterprise is enough, so the scalability is very good. In this architecture, the average programmer can program each other. It is not the same as our traditional approach, and data can be executed in chunks.

I introduced the situation of enterprise-class data center, the current ecological environment of our securities industry is like this, these days is relatively better, last year, our entire industry, 109 brokerage income is greatly shrunk, the most nearly shrunk by 7, 80%, the average shrinkage of nearly 50%. The entire industry has not found a better mode of operation, in the process of development there is competition, the result is white-hot competition, to take discounts, die to fight the way.

Since 11, regulators have seen a more dangerous state, we have in 11 years of investment in the business, it changed the status of the industry, the business has two ideas, first, the product must be fully in-depth analysis of the management, the securities industry is not products, because the introduction of the product can be a coup in the securities industry profit model. I've included these parts of the product lifecycle from requirements to service tracking, which requires business management and technology-related support.

The other is customer lifecycle management, including customer exit, in the midst of our marketing process it is very important to one node of a lifecycle, a customer he said if he doesn't trade here, he has a lot of reasons, now we have a popular word in the online shopping, Word-of-mouth, If a customer is away from our service system, he gives us a good reputation and brings us potential customers, which is good for us later. The securities industry is also studying how to manage the customer lifecycle in particular, we see in the entire industry there is such an opportunity and challenge, the first is that our industry will be industrialized upgrade, due to the customer lifecycle management, the product life cycle, like the industry's production line, As long as our raw materials are not the same, the product is not the same, but the middle of the process is standardized, because of the standardization, to achieve the scale of the industry, which we from the industrialization of such an angle to apply it in the financial industry so led out.

The second is agile services, a lot of time IT staff, we said that the business staff of the demand is not complete, the business staff said IT staff response is very slow, which both sides have a reason, but in the actual work of the promotion process, the end of the total one side mainly.

The new concept of the third operational dimension, which is also from the Agile service, our IT department is always a reactive type, the business unit to demand, IT department to do, this is a traditional it enterprise practice. In this day and age, especially like our network operators, he has to mention the concept of operations, it is the IT department is the initiative to promote, this is not a reactive way. This is the new concept of the 3rd movement.

The enterprise data center must be upgraded to a social data center. Because the enterprise-class data center, most of the data is derived from our business system, in fact it is a relatively closed, due to the development of the Internet, it is a lot of external data on our services, evaluation of our management, it's based on more , it gives comprehensive. This will inevitably upgrade our existing enterprise-class data centers to social data centers.

The other is the bi of Big data. We'll talk about it in a moment.

The industrial upgrading of financial industrialization requires us to have a standardization and industrialization. Agile services are related to it governance. We involve the enterprise-class data center, which includes the enterprise's Data Warehouse, which comes from our internal system and must be opened in the future.

Through our large data business intelligence, we put our customers in front of the entire lifecycle management and our products to the docking. We have in recent years to promote the financial industry in the cloud, a very sensitive topic is the data, the data on what to put on the cloud, our customer data are relatively private, sometimes we are in the process of advancing the cloud, sometimes too much security too stressed.

Let me give you an example of a time when we had no banks, most people put money in a jar buried in the ground, when the bank comes out, people feel it is not safe to put their money there, but often a professional service team to provide you with such a data service, in a sense, It may be safer than you are at home, which is why everyone's money is rarely put in the home, all in the bank.

In fact, cloud computing is the same, and at some level it's too security sensitive, in the process of implementing cloud computing in the enterprise, it must be a long-term planning process, which includes business management and technical aspects, which we proposed last year, especially in the securities industry, we involve a cloud ecological map, It has three forms, the first is the private cloud, the enterprise private cloud does not have to be placed inside the enterprise, I put on the outside, by the third party responsible for management, it is also a kind of corporate private cloud performance. We will use more in some large securities companies in the future.

This is the alliance private cloud, there are several organizations, for a common business development goals together, such a private cloud, for such organizations to serve, in the future this form can be for some small and medium-sized financial enterprises to serve.

This is the industry cloud, it may be more agile in the future, may involve industry cloud data, including public data, but the public data mining and analysis of different ways, in the future, some operators can provide these services, can provide relevant services in the industry cloud. This is our business in the Securities Industry cloud ecology.

In the middle of the enterprise cloud has a security control mechanism, simply speaking is internal defense, external plugging, audit, it for our enterprise cloud some of the risk problems I will not say.

In the securities industry, we designed a cloud storage architecture, as a result of such a technology, we have transformed the previous IT department's firefighting team into interactive form, from a standardized, scaled way, with the computer automated processing, we focus on innovative nodes.

Let me talk about the application of large data, the characteristics of large data, generally referred to large data, we are more concerned about is CE, mentioned large data is unstructured, in fact, large data concern its speed and its complexity. By creating a system that we set up months ago, we integrate our Customer system, service system, for example, in the process of developing a product, I can publish it to our new media to see the customer's response, which is not only the source of our securities companies to do business customers, A customer who does not make a transaction in our securities company, he may also make some comments and suggestions that he or she is in favour of, or oppose, so that it is more comprehensive. For a few months before the analysis of several processes is such a few parts, first of all to determine the subject, for the month before the analysis of such a current application, it is not possible to do a very wide set of a simple system, because the Chinese semantic analysis is very difficult, because the Chinese and English is not the same, English each word is divided, In addition to a word, there is no division between each word and word, if you are intercepting it, its position does not produce the same meaning is not the same, very difficult. So be sure to choose the best theme.

For example, the early release of a message, pharmaceutical problems, the news is negative news, this is a crackdown on the stock market process. In the capture of information to do some filtering, in our new media some advertising information, it must be done filtering. To do some preprocessing after the information is crawled, the better the work done, the higher the efficiency of the analysis behind us.

The establishment of the relevant model analysis, we see a film in the United States, is a special talk (English) growth of one, that is, for college students to paste the photos and then do some evaluation, in fact, we have a unique Chinese characteristics, market level is not very high, people by the influence of public opinion, If there is a certain stock on the trend of our market is very hot, and is to achieve a certain degree of relevance, it must be on our stock market is influential. Such a model you can use some good, very simple statistics can be calculated.

In this way, eventually we release it, this will form the basis of our internal decision-making, the external can form some data products, which is our data center in the process of building a application, and this application can gradually extend it to our trading system.

Everybody if the stock market knows, sometimes we look at the temperament of a stock has a different move, may be about to call to ask what is going on, this time we have been able to collect information on the site, such as Wenchuan earthquake out, we in the analysis system can be seen immediately, as long as you have a forward amount, At least when there is a negative effect, I can first sell a better price. But not necessarily a very good one, in this way the propulsion must be a systematic project.

In the midst of the application of large data, I think there are a few key points, the intelligent word breaker, it must be an accumulation of the process, it must have the function of learning, we in advance, found it and our realization is inconsistent, you must modify it, so as to make it more accurate.

The second is the storage, the large data can not be stored in every data, because you do not have such a large storage space, and you must follow the application, for example, we do in the application, we do not necessarily have all the data crawled to save.

Third, for structured and unstructured processing, transforming it into structured data, using a previously structured system can be achieved, and in the short term we can use our enterprise-class data center to make a gradient.

From product to service tracking, we can form our data products to provide our customers, can provide relevant basis for customers, can help him make some investment advice.

Together, in the process of cloud computing, it is a gradual process, so that enterprises should be their own information to carry out the relevant norms and a number of standardized carding, to find suitable for our own practical application of a framework, and gradually with our current popular technology and the way forward to combine, To achieve a multiplier effect. This is what I share with you, thank you.

(Responsible editor: The good of the Legacy)