Performance Simulation of Runtime Reconfigurable Component-Based Software Architectures

摘要

Architectural runtime reconfiguration is a promising means for controlling the quality of service (QoS) of distributed software systems. Particularly self-adaptation approaches rely on runtime reconfiguration capabilities provided by the systems under control. For example, our online capacity management approach SLAstic employs changing component deployments and server allocations to control the performance and resource efficiency of component-based (C-B) software systems at runtime.In this context, we developed a performance simulator for runtime configurable C-B software systems, called SLAstic.SIM. The system architectures to be simulated are specified as instances of the Palladio Component Model (PCM). The simulation is driven by external workload traces and reconfiguration plans which can be requested during simulation, based on continuously accessible monitoring data of the simulated systems. This paper demonstrates SLAstic.SIM including a quantitative evaluation of its performance.