Information department

Cost center --> profit center

In the past, date were stored in the information department and only used as records, and the enterprises only treat the data as a sort of appendage produced in system. However, through accumulating millions of these kinds of data, enterprises can better understand customers’ needs and be able to find new business based on the data. Storing data is no longer a necessary behavior that was only spending money, but it is helping the enterprises to find new opportunities to to find new businesses and make profits.

Strategic direction

Business driven --> data driven

In the old days, lots of enterprises make strategies for their enterprise based on their experience and subjective evaluations, which means that they were lack of deep-level analysis of customer, business, marketing and competition. In the age of big data, enterprises started to collecting and analyzing data from inside and outside of the their enterprise. Through mining these data, the enterprises can predict the market, and make strategies for different situations.

Decision making

Based on experience --> based on data

Decisions were usually made by people’s experience and sometimes instincts, however, big data changed all of this. The risk of decision making is no longer unpredictable because the data can speak for themselves. By data analyzing, the future can be predicted.

Facing the age of data, financial enterprises should covert their old business mode to data-driven business mode by improving their data operation system and implementing big data operation center. In this way, they can start their operation optimization, management promotion and risk control to improve their core value.

Operation optimization

Use the user data to comprehensively improve the operational efficiency. The user data includes customer portrait, precision marketing, product optimization, public sentiment analysis, market and channel analysis.

Management promotion

The management should be guided by building value contribution and input-output. Through performance assessment, leader’s cockpit, management accounting platform and other applications, the enter price can achieve refined management.

Risk control

Use multidimensional safety judgment, finer-grit modeling and forecasting to realize applications such as SME loan evaluation, real-time fraud transaction analysis, anti-money laundering business analysis, and etc.

Figure 3: Architecture Diagram MPP Data Mart of vividime

ETL layer:

PC servers are used as ETL front-end processor to clean, transform and load data.

Real-time offline analysis system:

The technical architecture uses Hadoop data center to do distributed storage and some data model calculation based on the analysis needs. It can dig and analyze large-scale batch computing tasks in low timeliness. When data size increases, scale-out can be done to the system to storage more structured and unstructured data.

Real-time online analysis system:

The technical architecture uses vividime high performance engine VooltDB for the distributed data market. VooltDB supports high concurrency and high availability. Every data market contains data details for light modeling on the basis of the topic. Data are distributed stored in the nodes and backed up at the same time. They are compressed as the way of columnar storage, labeled and stored on disk. When query computation required, the memory calculation is used to calculate data, then each node will calculate at the same time. At last, the result will be presented at the application layer.

Application layer:

vividime agile BI provides self-service analysis to visually display the data from offline and online analysis platforms.Both BI users and IT developers can access the BI system through a major browser, and users can access the system via a mobile terminal too. BI system provides self-service report design and data analysis, and supports system monitoring, multi-level authority management, multi-dimensional data analysis and other functions as well.

vividime give our customer a solution that do not need to purchase expensive minicomputer to support high concurrency to support mass data calculation, data analysis, or business development, but only need to have several ordinary PC Servers to set up clusters and construct highly cost-effective analysis platform. Our technological architecture also allow scale-out for upgrading the data volume on both data market and BI front.

In the data layer, our system use lightweight model to import data, which means that when the models are made, the users only need to connect the data source. In the application layer, our BI provides a very simple experience. In the datasets making, vividime allow users to use self-service modelers to join, union, and clean the data based on drag-and-drop actions, and the SQL will be automatically made by the BI. In the dashboard making process, the users also will use drag-and-drop to make the dashboard with our over 80 components. Easy and convenience are the experiences that vividime gives to the customers.

Both of the online and offline systems are distributed architectures. Distributed storage will have backup, and it help the enterprise to prevent data lost caused by disasters. When one machine goes down, the other machines will have the backup data and automatically continues all of the calculation.

VooltDB supports high concurrency. Our VooltDB is a data engine that is designed for vividime, and it supports second responses for data size of tens to hundreds millions. VooltDB uses columnar storage to store data, so it uses high speed memory computing, which is very suitable for real-time analysis of massive data.