Abstract

An agent-based simulation model was built for studying resource management in the clouds, focusing on how service placement strategies, service auto-migration, and server consolidation affect the overall performance of clouds of homogeneity and heterogeneity, with regard to power consumption, resource utilisation, server-level agreement violation. Major cloud elements were modelled as autonomous agents with their characteristics encapsulated. The complex relationships between components are no longer to be hardcoded rather than realised through asynchronous agent interactions. In the evaluation, performance of several resource management algorithms was studied using statistical methods, and accuracy of server power consumption models were evaluated.