基于相对资源量的集群系统的研究与实现

Design and Implementation of Cluster System Based on Relative Resources

在中小银行核心系统中,批量业务非常容易使系统资源消耗达到最大,重要表数据也容易被锁表,严重影响日常实时交易,是核心系统性能的重要瓶颈之一。阐述利用事务联机分析中高效的ETL并行计算架构,根据中小银行核心系统大部分批量业务处理都是非实时性情况,研究设计基于相对资源量的集群系统,将批量业务所需的数据导出到集群系统中处理,处理结束后再导回核心系统中,极大减轻批量业务给核心系统带来的压力,增强核心系统稳定性与延长使用周期。集群系统中动态负载平衡算法,使用相对资源量概念,给节点预定拥有资源量,给作业预定消耗资源量,负载平衡器根据节点剩余资源量与作业消耗资源量动态分配作业。

In the small and medium-sized banks in the core system, bulk business is very easy to reach the maximum system resource consumption, the important table data easily lock table, serious impact on the day-to-day real-time trading, that is one bottleneck of the core system perfor- mance. Proposes cluster system based on relative resources design scheme which consists of its logical architecture,deployment and implementation etc. The cluster system greatly alleviates the pressure of bulk business for the core system, and enhances the stability of the core system and extends the life cycle.