Automated Health-Assessment of Software Components using Management Instrumentatio

Software components are regularly reused in many large-scale, mission-critical systems where the tolerance for poor performance is quite low. As new components are integrated within an organization´s computing infrastructure, it becomes critical to ensure that these components continue to meet the expected quality of service (QoS) requirements. Management instrumentation is an integrated capability of a software system that enables an external entity to assess that system´s internals, such as its operational states, execution traces, and various quality attributes during runtime. In this paper, we present an approach that enables the efficient generation, measurement, and assessment of various QoS attributes of software components during runtime using management instrumentation. Monitoring the quality of a component in this fashion has many benefits, including the ability to proactively detect potential QoS-related issues within a component to avoid potentially expensive downtime of the overall environment. The main contributions of our approach consist of three parts: a lightweight component instrumentation framework that transparently generates a pre-defined set of QoS-related diagnostic data when integrated within a component, a method to formally define the health state of a component in terms of the expected QoS set forth by the target environment, and finally a method for publishing the QoS-related diagnostic data during runtime so that an external entity can measure the current health of a component and take appropriate actions. The main QoS types that we consider are: performance, reliability, availability, throughput, and resource usage. Experimentation results show that our approach can be efficiently utilized in large mission-critical systems