An approach to reduce QoS monitoring overhead in ESB

Abstract

Since the Small and Medium Business (SMB) markets is growing, service-oriented architecture will play a crucial role in the SMB IT market. In this market the ability for better integration, increased flexibility, and cost reduction in development by reuse of existing services must be considered. Because of the service oriented architecture's dynamic nature, increment the number of clients and high volume transferred information between service provider and service consumer in business processes, Quality of Service (QoS) monitoring is needed. Services can be monitored at run-time to check whether they comply with their contracts or not. Monitoring can be done in different places such as provider side, client side and third part like Enterprise Service Bus (ESB). In the approach presented in this paper, the monitoring module was placed in ESB. Despite the monitoring place, QoS monitoring module is invoked at every time of service calling. Because of other methods running, overhead is inevitable. Specially monitoring overhead effects on the service quality parameters such as response time. The main objective of our approach is to reduce the QoS monitoring overhead using Time Series Forecasting in Neural Network. At the end, the performance of approach was evaluated through simulation on a case study in Tehran Metro. The experimental results show that this approach had low overhead in monitoring module.