Home >

Zhang Dong: The Ultimate Goal Of Data Is To Serve People.

2014/5/22 19:37:00 43

??? 日本語 Fran?ais English 中文

Zhang Dong concluded: first, value density; secondly, data often have strong independence and sharing obstacles; finally, big data must have more sources, cross industry and cross domain data collision, so as to truly call it big data.

to talk about the problems, Zhang Dong put forward: first, the problems of collecting and summarizing areas; two, the problem of data quality and the disunity of data formats; three, the restrictions on some policies or management systems, resulting in a lot of data can not be unified.

below is Zhang Dong, director of cloud computing technology group of the wave group:

Zhang Dong: I am very happy to have the opportunity to share with you all experts and guests here some ideas and developments of cloud computing in the next few years.

in the front, we also talked about many waves of cloud computing concept, we talked about the industry cloud, we in the industry to promote the development of China's cloud computing, how the industry cloud landing, how to step by step toward big data.

this year's theme is cloud computing and big data to promote smart China.

What I am going to talk about today is the challenges we have seen from the industry cloud to the big data in the industry, the promotion and application process of the whole big data, and the wave in this regard has launched our big data solution for more industries.

this is a development path that we have been talking about for several years. From the very beginning, we talked about the cloud computing strategy of the wave. When we talked about the strategy of cloud sea, we just talked about cloud computing. The first step was to gather our resources together. On the one hand, our cloud computing has the advantages of cloud computing, which can reduce resource consumption and improve our resource utilization.

However, there are more ways we can integrate these resources into the business and data they carry, and integrate them, and further enhance our economic and social services through higher data mining and utilization.

so in this sense, we think that from cloud computing to big data, it is actually a complete path from which we gather resources to data utilization.

Especially now, we have talked about cloud computing for several years, and many systems may have been built. How can we really achieve the integration of data in a cloud and break the barriers between data? This is probably the most important work to be done at this stage.

therefore, for data, if these data are like the biggest problem that China often talked about before is the isolated island of information, all data may be scattered in different places, different formats and different mastery.

we think that from the data point of view, it may be almost the same as cloud computing. For example, cloud computing we need to do physical centralization, business concentration and data integration first. For data mining, it is also a small number of people who have data at the beginning. I may have my own data. I will mine my own data, and gradually how to achieve data interaction between different users, and then to the end, in the final sense, all the big data.

the big data is not only large in scale, but more in that I come from different sources. I may have generated it myself, or there may be many outside, such as the Internet, or from the media.

three stages of the development of the industry. Data mining and application from the organization to the data sharing and application of different business modules within the organization have been applied to cross industry data.

it should be said that big data has been talked about for several years now, but our feeling is very hot in the Internet field. Maybe in the field of science, a lot of professional data people are also very enthusiastic, but on the contrary, in some industries such as public security, industry and Commerce, tax and so on, many of China's important economic departments and industry departments, in fact, data applications are still facing many problems.

What are the characteristics of a href= "http://www.91se91.com/news/index_c.asp" > industry data < /a >? We simply summarize some characteristics which may be slightly different from our general data. The first value density is density.

When we talk about several V of big data, we will talk about density of value, because its data sources are complex, quantity is large, value density may be relatively low, and better processing technology is needed.

And we think that in the industry, its value density may be higher, because its data collection is often purposeful, for example, in our industry and commerce or tax, its data may be data collected for its business.

In view of these data mining, in fact, can produce greater and better value.

we know that to do anything, we may be willing to find the best place to deal with, and can quickly see the benefits, such as water always looking for the fastest place to flow.

Similarly, when looking for such an opportunity, we also feel that in the industry, we can better find some opportunities in this industry by mining big data, find some rules in it, and better serve our economy.

second is not a good idea. We think this is a bad feature for sharing and mining big data.

It is in these industries that these data are often highly independent, and there are many obstacles that you want to share.

You may all know that these data share it, and it may be beneficial to deal with it comprehensively. However, for various reasons, there may be reasons for management system, or there may be motives for these people, or the reasons for not having the motivation to do things. The proportion of data sharing is very low.

Some of the numbers we are seeing now may be less than 10%, or even lower in some places, and this phenomenon is not only in some industries that I have just referred to, but also in many of our enterprises. For example, financial data and manpower data may not be shared or shared.

There is a very important problem in , because the owners of these data are often different. Because of the different owners, it is difficult for you to connect these data together.

If you want to solve some of the owners' concerns, for example, if I put them together, will I divulge secrets? If I put them together, will there be some things that others should not see or tamper with? How can they ensure their safety?

the third characteristic is that we think that big data is different from previous data analysis. That is, it emphasizes the data generated by us in the industry and even on our business occasions. The comprehensive analysis is called big data. If we simply run out of the internal ERP or the financial data in the financial statements, this may be ten years ago or 20 years ago, and many people can do it.

Big data must have more sources, cross industry, cross domain data collision, in order to truly call it big data.

the last one is to say that in these industries, the final purpose of data, from the previous industry cloud to the back industry big data, is to say that data must eventually be served for people, and that everything must be pformed into data service. Otherwise, if you analyze and excavate these things, and finally put them in your database, if you do not share the results with others, or do not produce services for more people, then the construction of cloud and the construction of big data systems may not be so great.

however, such an internal data is pformed into a service. In the process, these data should not be disclosed, or how to prevent it from being tampered with outside, how to do the reliability and availability of data, and how to do some quality of service in this area. These are all some different features which are different from those before us in the development process of the big data of the industry.

What are the problems facing p? First of all, we begin to gather big data from a data collection. Until the end, we analyze it and make decisions, and show it to such a process, including collection, storage, analysis, visualization, and ultimately for decision-making.

At every stage, there are still many problems in fact. If we want to extend the big data to more fields and enable more industry users or traditional users to use information, we must solve this problem.

the first is the collection and < a href= "http://www.91se91.com/news/index_c.asp" > < /a > field.

Because in the past two years, we have been looking for many users in the name of big data. Many users have heard that the big data is very good. After they have finished the budget, they want to make big data. They are also very enthusiastic.

But if you look at his system, the first question is, where is the data? A lot of people want to do things. Actually, there is no data that he wants to do. The first problem is data collection. Maybe a lot of data are generated, but for him, he has no effective way to collect it into the system.

Of course, this collection may be a technical reason or a management cause.

span style= "font-family:" Song body ";" font-size: ";" "" "" > "< < >", "song";

< --EndFragment-- > second. Even if he has collected some data, the sources of these data are very complicated. Maybe they are generated by himself, for example, he collects himself, uses machine to collect, and uses people to collect, and some of them are borrowed from others. Then you have no way to ensure that these data are all good, or are very suitable for doing his later business.

That is to say, the quality of data is a big problem, and data format is also a big problem.

finally, the restrictions on some policies or < a href= "http://www.91se91.com/news/index_c.asp" > management system < /a > have caused a lot of data to be ununited. An example may be mentioned later. It is a typical department in the country, that is, public security. We all think it should be a department, but later it is found that N is a multi department. All the data in it are complementary and connected. This phenomenon is not only in such an industry. We have seen such problems in many industries.

the first problem to be solved is how to put these things together and solve some problems of technology in the collection, and how we can do the standard and how to standardize it.

second questions, this data is gathered together, and there will be a problem of storage and management.

In fact, many people say that I have a good collection method, I have many cameras and sensors, but these things are very difficult to come back, the background is not so large storage capacity, in two days will have to roll it once again.

In this case, with the increasing size of data and the increasing complexity of data types, there may be pictures and videos, which can be placed in the database, and can not be placed in the database. Now many of them have stored many things in the database, and the database can not be replaced slowly, because it is too big to put down.

there is also the problem of sharing and privacy in the unified storage process. We will make a request with more than one user.

A unified big pond, the data are placed in front of me, and now I want to put the data in his place. How do you guarantee that the data I put on him is safe? You must give me the means and means. Of course, many users are hearing the traditional means, such as mandatory access control, encryption and so on, but in this way, some traditional security means should be said that this is a problem that has been being explored after the cloud computing mode has been generated, and it is also a problem that has not had a very good answer. That is to say, after I gather, how to prevent infiltration between users, and second, how to prevent the backstage administrator, whom I have never seen before, I have to trust him. Now we have built one.

the third problem is the problem arising from the analysis process. From the front contact, the analysis process is simply two.

First, people who know business can not write programs. People who write programs do not understand business. How to pform these knowledge into knowledge of computers is actually a very troublesome thing.

Just now, many experts have talked about some experiments done in some big enterprises to do face pattern recognition, including voice, video and audio, and how to train machines to work like human brain.

But in many professions, there are still many more professional knowledge.

For example, the administrative field may depend on people. For example, now our medical treatment, of course, experts say that medical treatment has been moving forward, but it is also the same in the medical system.

It is not through the computer network, we can take our own symptoms and test sheets to enable computers to help people see a doctor, and solve the medical problems that we are faced with many times. When this is another topic, how can we pform people's experience into computing emotions? This is what many people will be faced with. It is difficult for people to find new applications. In fact, the written applications can also be done by OA. It may be very complicated to do a real business together with his business, and now there are many algorithms or deeper levels of things that have not been broken through, and have not been resolved. When we were discussing with our colleagues a while ago, we were also talking about this problem.

Second, that is, this application is too complicated. In the past, we all used data technology, and later we used data warehouses. Then one day, we all said that Hadoop is a good thing. We all have Hadoop.

Is not all applications suitable for Hadoop? In fact, it is not. From the original database to Hadoop, it is not possible to cut back.

How to find a variety of technologies suitable for application to solve its fundamental problems is also a challenge now.

is the problem of visualization at the end. Now many people are accustomed to using pad and mobile phones to show these things. It is also a big problem.

before we talk about some challenges, let's talk about some of the work done by the wave in this respect.

First of all, as we have just said, big data is also a step forward in the wave of the whole cloud computing strategy. We start from the industry cloud to help users to do computing, gather resources and analyze data. This is due to our current cloud computing solution.

The wave is still focused on our data center, from the construction of the bottom computer room to servers, to storage, to the system software we provide the foundation, and to provide advisory services, planning services and so on to provide users with a large set of cloud computing and big data system.

in the past year, the wave has also introduced many new products in cloud computing, including our modular data center, and the Internet Oriented high-density data center. Now it has a high share in the Internet market, including our industry oriented big data area and so on. If you are interested, you can take a look at the presentation in our exhibition hall.

just mentioned that we need to take a step by step to plan cloud computing for a user. We think that the previous step is not to jump from zero or one to three, but gradually start from building cloud to gather data, and we also provide users with a complete set of construction plan consultation.

specific to big data, the key to our proposal is integration. How do we understand integration? We think it is divided into three parts. The first is that all the processing processes we mentioned before can be solved by using such a system from collection, storage, analysis and visualization.

Second, give full play to the advantages of the wave in hardware, improve the overall performance through hardware and software integration, including the acceleration of hardware and the ability to design large memory computing, and improve the performance of the whole big data processing.

Finally, the integration of solutions, facing different industries, like this year, we released an integrated machine for the financial industry. In the future, we will release an integrated machine for public security industry to make the overall solution.

finally, we will share with you some examples of solutions. This is actually our case. We are still in the process of construction. This is a whole case of provincial public security from cloud to big data.

These questions mentioned earlier have just been mentioned. For example, we also think a government department may lead a sentence. Everyone can handle everything well. Actually, everything is not easy to handle.

There are many things, for example, data sharing. Before that, the Interpol, economy, household registration, customs and immigration are all separated. Data sharing is very bad before that. Business system duplication is very serious. All systems are running on a single machine, and data processing capability is poor.

this is what we have just said. Each system corresponds to a set of hardware, corresponding to a set of databases, corresponding to a set of independent data. When you want to do the application, you can apply it to the application, and you may not be able to run on that.

IaaS layer, to the middle of the wave based open big data processing platform, we can be understood as a PaaS, but we are still a preliminary data sharing and connectivity platform, it will focus on all its IT systems, and gather all its data together. In fact, its original application, the business system can not move. As long as I run on top, visiting the special data of that place, it will form a new platform, which will share all the internal data, and bring in the data in the outside. It can carry out multi point collision, realize more problems that it can not solve before, or solve problems that can not be solved quickly, thus forming a whole platform of cloud computing and big data. We plan for him from the bottom, including the bottom.

this is our specific plan. This is a physical structure. It combines the platform of the whole province by means of unified monitoring and management. This is a way of processing data in big data, for example, it has very traditional database. For example, many people's household registration information may be in the database, but it has a lot of audio and video data. It uses a variety of platforms, including databases, new platforms like Hadoop, to build a unified platform.

above is a data sharing platform of our wave, also called IOP, which uniformly extracts all the cloud data at the bottom layer, unifies processing, and displays it on the above, thus forming a big data solution for public security.