The entire series of articles describes the data warehouse from the bank data warehouse architecture, ETL, models, data management, and several aspects of the application, so that we can have a summary of understanding on banking data warehouse.
The foregoing review:
" What system architecture bank data warehouse is? See this is enough . "
" Depth Analysis | article read data architecture banking system ."
" Banking Data Warehouse are like this, article for "
But not too in-depth introduction in design, technology subsystems, follow-up will continue to add. As the last one of this series, talk briefly about some ideas for data warehouse development banks.
Bank data warehouse as a data center, in this era of big data also played an important role, and that with the evolution of the system architecture of the banking business, especially now that the current banking systems shift to a distributed architecture using micro services in, increase throughput and efficiency of the system, to adapt to the characteristics of the Internet high-volume and high number of concurrent users. You may have the following changes consequent:
1. decentralized data:
After the loan before the system includes a pre-loan application, the loan amount in the nucleus, there are many post-loan management and real-time query, in the micro-service architecture, a system may be divided into a loan before the loan, the amount of nuclear, loan, front-end (H5, app , pc) and other business systems, but also calls text messages, multiple signatures associated systems, customer information systems. A business process that data will be distributed to each subsystem.
2. The amount of data growth:
Bank's own Internet through a variety of channels (phone, public number), as well as through cooperation with the Internet platform, open API and so will play the role of the Internet in the background that the user may feel the existence of the bank, but in the use of banking services as many banks currently own two, three types of accounts offer the ability to Internet companies, customer accounts opened in the Internet is in fact the other banks opened. Therefore, the account number and the amount of the bank's customers will be rapid growth in trading volume is followed by the outbreak.
3. Data diverse applications:
With the development of big data and AI technology, data applications will grow substantially. While statistical data and real-time applications demand will more appear, wind control, marketing, investment decisions, and other anti-fraud services model, such as real-time monitoring and real-time traffic warnings, at present FLINK, SPARK STREAM real-time data processing platform development is rapid.
Big Data software Finereport made double 11 big screen
Decentralization and diversification will inevitably bring data applications data needs blowout, increase the amount of data scalability and performance requirements of the technical platform will be higher, the data warehouse also need to gradually evolve the architecture and functionality to suit business development needs:
1. Technology platform technology platform for higher performance and greater storage will appear
HADOOP appeared open source platform, reducing the technical barriers and cost data warehousing and big data platform, but the internal (google, Facebook, Ali, Tencent, Huawei, etc.) in terms of the commercial version or the ease of use and efficiency of manufacturers optimized version advantage, both sides will learn from each other follow-up, especially in some giant contribution to the open source community will encourage more powerful data processing technology platforms appear.
2. Data warehouse multi-cluster
As the amount of data as well as increased demand for applications, in order to reduce coupling and increase flexibility in data warehousing will be distributed to different functions and multiple clusters without clustering technology platform may be different, such as basic data area, each data mart, real-time applications, historical data, unstructured data, etc. may all build a cluster alone, so rapid exchange of data between the clusters will be more demanding. There may be subsequent emergence of new technologies and the like from the aspects of the underlying shared memory or copied.
3. Data Warehouse universal skills
Also with the data fragmentation and increased applications in order to improve efficiency, there will be multiple teams to develop on data warehousing, data warehouse will be a basic platform and basic skills, resources need to be isolated, while the need for data the functional components of the warehouse, tools, standardization, reduce learning costs in the bank or company-wide promotion, improve development efficiency.
4.AI platform and data warehouse technology platform integration
AI will follow gradually become a common feature, hadoop ecology now also supports machine-learning component, such as a mahout, sparklib, but professional and AI platform is still a gap algorithms, function and performance, AI platform also supports data as hadoop processing platform, so both will follow the gradual integration, more intelligent data processing technology platforms appear.
The future is not far away, the world's science and technology really can be a tremendous pace. As a data warehouse development, product manager or managers, at the same time doing routine work requires continuous learning new skills in order to have technical reserves to provide efficient data services when new data requirements arise.
Also you need to be familiar with the data content and business processes behind the data, look for optimization point behind data service or product. Begin with the End, promote the development of business and products.
