Research Article
Optimal virtual machine scheduling with Anvik
@INPROCEEDINGS{10.4108/icst.valuetools.2013.254403, author={Andrea Sansottera and Paolo Cremonesi}, title={Optimal virtual machine scheduling with Anvik}, proceedings={7th International Conference on Performance Evaluation Methodologies and Tools}, publisher={ICST}, proceedings_a={VALUETOOLS}, year={2014}, month={1}, keywords={cloud virtual machine allocation scheduling admission control markov decision process}, doi={10.4108/icst.valuetools.2013.254403} }- Andrea Sansottera
Paolo Cremonesi
Year: 2014
Optimal virtual machine scheduling with Anvik
VALUETOOLS
ACM
DOI: 10.4108/icst.valuetools.2013.254403
Abstract
In Infrastructure-as-a-Server (IaaS) clouds, the provider has to decide on which server the virtual machines (VMs) requested by the users should be provisioned. This is an online scheduling problem, in which incoming VMs, typically with different resource requirements, have to be scheduled to one of several heterogeneous servers with limited capacity. For the provider of a public cloud, the objective is to maximize profit, the difference between the operating cost of the servers and the revenue due to running the VMs. In some cases, it might be advantageous to perform VM admission control and reject low-profit VM if low-cost servers are unavailable. We model this problem as a continuous-time Markov decision process and present a tool, Anvik, for the computation of the optimal scheduling and admission control policy. Anvik is released as open-source.