Skip to Main Content
System execution traces are useful artifacts for capturing distributed system behavior and validating quality-of-service (QoS) properties, such as end-to-end response time, throughput, and scalability. Although system execution traces assist in validating QoS properties, system execution traces are very dense. This is because system execution traces capture a system's complete behavior and execution in its target environment, which can consist of many components deployed across many hosts that execute for long periods of time. It therefore can be hard for distributed system testers to effectively analyze different views of QoS properties using system execution traces and/or pinpoint performance bottlenecks that need to be resolved. This paper provides two contributions on validating distributed system QoS properties. First, this paper presents Aspects for System Execution Traces (AsSET), which is a join point model for applying aspects to system execution traces and enabling context-based analysis of system execution traces. Secondly, it shows how applying AsSET to a representative enterprise distributed system can improve QoS analytical capabilities for distributed system testers. The tools and techniques discussed in this paper have been realized in an open-source system execution modeling tool named CUTS, which is currently used on several industry-related projects.
Date of Conference: 23-25 Aug. 2010