Design And Implementation Of Batch Processing System For Banking Business Based On Microservice Architecture Microservice Architecture

With the vigorous development of social economy and Internet finance,people's demand for financial products has also undergone profound changes.In order to meet the diversified needs of customers for financial products,each bank has rapidly expanded all the types of business on the basis of the original IT system.However,most of the traditional banking system is monolithic structure,and all business functions are coupled together,so any part of the function modification will have an impact on the entire system.The continuous increase in system functions will cause the entire application to become larger and larger,and subsequent maintenance will become more and more difficult.With the rapid growth of business data,the traditional batch processing systems for banking business are becoming more and more inapplicable:1.The traditional development method of batch processing business only focuses on the realization of functions and lacks a unified development specification,so the business logic is chaotic and the development is low-efficient;2.The serial batch processing method adopted by the traditional business batch processing system cannot realize the efficient processing of increasing business data.3.The traditional business batch processing system relies on manual operation for batch task submission,and it becomes unrealistic as the business process becomes more and more complicated.Considering that the microservice architecture has the characteristics of low coupling between services and services can be extended on demand,this thesis designs and implements a batch processing system for banking business based on the microservice architecture,aiming to solve the aforementioned problems faced by traditional batch processing systems.Firstly,in order to unify and standardize the development of traditional batch processing business,this thesis adopts the idea of componentization,designs and implements a unified and standardized development model of batch task based on the characteristics of the bank's batch processing business.At the same time,the business process orchestration function is designed and implemented.The bank's business developers only need to implement specific business logic according to the task model,and then arrange the tasks in accordance with the established sequence to complete the batch development quickly.Secondly,in order to improve the efficiency of the traditional serial batch processing,this thesis focuses on the segments of the batch processing business data based on the key field of the database table,and then these multiple segmented data are distributed to different task executor nodes for simultaneous batch processing.Through this multi-node parallel processing method,the efficiency of batch processing is improved.Finally,in order to automatic execute operations of the traditional batch tasks,this thesis designs and implements a batch task scheduling center based on the time wheel scheduling algorithm and database polling strategy.Developers can perform unified control and automated scheduling of batch tasks through the task scheduling center.The system has been successfully run in many banks.Practice has proved that the use of this system improves the development efficiency of the bank's batch processing business,and realizes the automated scheduling of the bank's batch tasks and the distributed parallel batch processing of massive data.