Custom Windows Performance Counters Monitoring Mechanism for Measuring Quality of Service Attributes and Stability Coefficient Service-Oriented Architecture

Service-Oriented Architecture (SOA) has been widely used for different types of systems as their underlying architecture. The most popular technology that implements the SOA is web service. When several web services provide same functionalities, Quality of Service (QoS) of web services turn to be an important issue. In this study, monitoring is used in order to measure QoS attributes of web services in SOA. Several monitoring mechanisms have been proposed. Windows Performance Counters (WPC) is one of approaches for monitoring services at provider-side. However, WPC monitoring approach has a limitation and it can be employed just for WCF services. Moreover, predefined system counter values do not map to QoS values properly. In this research, a new provider-side monitoring mechanism which is based on Custom Windows Performance Counters (CWPC) is proposed in order to overcome current limitations. CWPC will be set to measure QoS attributes of web services such as response time, throughput and reliability properly. The results of CWPC monitoring are useful in taking decision in adjusting suitable monitoring interval for the system. Additionally, the result verifies that CWPC is an accurate monitoring approach for measuring QoS attributes. Besides that, this study also focuses on variability of QoS values which are obtained by monitoring of web services at different service invocation time. QoS values are variable and service consumers may experience various QoS values due to the fact that web services run in a distributed, dynamic, and unreliable environment which makes them exposed to faults and failures. In this research, a new Stability Coefficient is introduced to measure stability of a service based on historical QoS values that were obtained by monitoring the web service. Such a measure enables service consumers to find a stable and trustable service based on QoS attributes and it can increase consumer’s satisfaction. In this study, the Stability Coefficient is defined based on an average of different QoS attributes of service stability. The results confirm that the proposed Stability Coefficient is a proper criterion for determining stability of services in terms of their QoS attributes and a stable service with less QoS values variation has a high Stability Coefficient which may lead to more satisfaction to service consumer