Abstract
Cloud offerings serve workloads with diverse needs in terms of compute, memory, storage and networking. In this work, we address the problem of workload-aware virtual machines (VMs) placement in Infrastructure Cloud. The goal is to co-locate VMs with heterogeneous workload characteristics to optimize performance and reduce energy consumption. We design and implement the proposed placement technique on top of the popular Cloud Infrastructure Service, OpenStack. Different workload combinations for a server are shown to result in different energy consumption and performance. We demonstrate through real-world application benchmarks of varying characteristics that the proposed workload-aware placement strategy can achieve improvement in average performance and energy savings of up to 20% as compared to OpenStack default VM scheduler with minimal overhead.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
