Based On Cluster Technology, Banking Transactions Middleware System

Resent years, all kinds of banking middle agent operations have been developed rapidly. But existing banking agent software with C/S model results in complicated banking network, low system efficiency, hard maintenance, and bad system scalability. On the basis of present situation of banking, this dissertation puts forward a banking business middleware, a whole resolution scheme of system platform based on three layers model and clustering technology.Considering present network layer's complicated and disordered scenery leaded by communicating among various application with different communication protocols and technologies, this dissertation puts forward and implements shield of complicated network communication with standard protocols of TCP/IP, UDP in middle layer and Socket network programming technology, which can greatly alleviate application software personnel's workload, ease its technical difficulty, and enhance system universality and transplantability. Aimed at situation of overload of some host computer and low or empty load of others in banking host computers group, the dissertation adopts the clustering technology and brings forward a network dynamic load balance algorithm grounded on node load, that is, different assignments is dynamically distributed to low or empty load and assignment-suitable node by load balancer. Based on research on clustering technology and signal mechanism, this dissertation also advances and realizes a method to monitor and manage nodes in clustering; meanwhile, to avoid system efficiency decline and then breakdown caused by simultaneous running of vicious attack or disposal of large quantities of transaction, this paper advances a control algorithm based on node valve to effectively resolve this problem.Practice proved that transaction middleware for 'banking application -with clustering technology can effectively resolve the problems of heterogeneous network communication and heterogeneous database interaction, reduce workload of new application, simplify system complexity, and equipoise assignments, as a result,optimize the whole performance index and enhance the whole system efficiency.