Heuristics for Scheduling Virtual Machines for Improving QoS in Public Computing Utilities

Computing utilities based on virtual machines (VMs) migrating among distributed computing resources are becoming increasingly practical and emerging as a viable infrastructure for outsourcing computer services. Here, we present a large scale distributed computing architecture named public public computing utility (PCU), where nondedicated public resources are augmented with privately owned dedicated resources to obtain higher levels of quality of service (QoS) for compute intensive applications. Virtual machines are used to isolate foreign applications from the host systems and to facilitate seamless migrations. We have proposed several heuristics for scheduling VMs among the resources in PCU. Extensive simulations we performed indicate that the proposed heuristics are able to improve the QoS delivered by the PCU by utilizing the public and private resource combinations while reducing the VM migrations.