Research Article
A VM Vector Management Scheme for QoS Constraint Task Scheduling in Cloud Environment
@INPROCEEDINGS{10.1007/978-3-319-38904-2_5, author={Kyung-no Joo and Seonghwan Kim and Dongki Kang and Yusik Kim and Hyungyu Jang and Chan-Hyun Youn}, title={A VM Vector Management Scheme for QoS Constraint Task Scheduling in Cloud Environment}, proceedings={Cloud Computing. 6th International Conference, CloudComp 2015, Daejeon, South Korea, October 28-29, 2015, Proceedings}, proceedings_a={CLOUDCOMP}, year={2016}, month={5}, keywords={Cloud computing Scheduling workloads SLA}, doi={10.1007/978-3-319-38904-2_5} }- Kyung-no Joo
Seonghwan Kim
Dongki Kang
Yusik Kim
Hyungyu Jang
Chan-Hyun Youn
Year: 2016
A VM Vector Management Scheme for QoS Constraint Task Scheduling in Cloud Environment
CLOUDCOMP
Springer
DOI: 10.1007/978-3-319-38904-2_5
Abstract
To reduce operational costs in computing service, there have been many researches on resource utilization improvement. In cloud environment, virtualization technology, coupled with virtual machine migration, can improve utilization of physical machines by server consolidation. Cloud service providers will consolidate virtual machines in order to reduce the number of physical machines running, therefore reducing their operational cost. Capacity of resources used by virtual machines can be set by users who schedule their tasks, minimizing resource waste by underutilization. However, it is difficult for a user to find the optimal virtual machine with respect to the resource capacity in minimal cost. To solve this problem, cloud service broker is required between users and cloud service providers. Task scheduling in cloud service broker solves finding virtual machine with lowest cost while satisfying SLA. Previous methods using mixed integer programming have showed difficulties in complexity and as system got larger and more complex, they could not solve the problems effectively. In this paper, with preliminary experiment, we propose vector modeling on virtual machine types and tasks can be applied and used in VM management. The allocated computing resources for each task components showed low complexity in operation of VM managements and effectiveness in task consolidation.