轻量的银行大前置交易执行引擎及其实现

A LIGHTWEIGHT TRANSACTION ENGINE FOR FRONT BANKING APPLICATIONS AND ITS IMPLEMENTATION

新业务是银行业的利润增长点,为了尽快推出新业务,可以采用改动银行核心系统的方法,但是该方法实现困难而且容易引入错误,银行大前置系统应运而生。基于银行大前置系统,可以快速开发和投产银行的新业务,拓展银行的收入来源,提高客户满意度,进而提高银行的竞争力。目前国内银行业使用的大前置系统具有技术复杂、扩展性差的缺点。为此提出轻量的新一代银行大前置交易执行引擎设计方案,该交易执行引擎具有简洁的系统架构,支持高度的系统扩展性;通过可插拔模块的方式支持各种网络协议,实现与各类第三方系统的连接;该执行引擎把Session管理交给数据库系统,极大简化了核心引擎的管理负担;在私有变量池、元信息共享的基础上,利用新的交易执行策略,充分发挥多核技术优势提高系统性能;在数据展示技术方面,利用基于哈希算法的XSLT技术完成展示层的渲染,支持大量用户通过B/S模式对系统进行访问。在与传统交易执行引擎的性能比较实验中,该系统获得了23.2%的性能提升。

New services are the sources of profits for banks,it is difficult and error-prone to revise the core banking system to deliver new services quickly,therefore the front banking system comes into being.Based on the front banking system,new banking services could be developed and deployed online quickly,the income sources of the banks could be expanded,a higher degree of customer satisfaction could be raised,and in turn the advantageous competition position of the banks could be expected.At present,the front banking systems deployed by domestic banks have drawbacks of complicated technology and poor scalability.A design scheme of lightweight transaction engine for new generation front banking system is proposed in the article.It has a concise system architecture and supports high degree system scalability.It also supports various network protocols by using pluggable modules,and realises the connection with different kinds of third-party systems.The transaction engine delegates session management function to a database engine,in this way it greatly simplifies the management burden of the core engine.In addition,a new transaction strategy,which is based on private variable pooling and meta data sharing,is employed to fully exploit the technical capability of multi-core for improving system performance.On the aspect of data show technology,a hash function-based XSLT technique is used to complete the rendering of the show layer,which supports mass users accessing the front banking system via B/S mode.System throughput of a mix workload is improved by 23.2 percent over a traditional engine in our experiments.