Submitted version, 259.4Kb
Published version, 484.4Kb
Supporting information, 339.7Kb
Supporting information, 351.7Kb
Ko, R. K. L., Tan, A. Y. S., & Ng, G. P. Y. (2014). Time for Cloud? Design and implementation of a time-based cloud resource management system. In Proceedings of IEEE Seventh International Conference on Cloud Computing, Anchorage, Alaska, USA (pp. 530-537). Washington, USA: IEEE.
Permanent Research Commons link: https://hdl.handle.net/10289/9017
The current pay-per-use model adopted by public cloud service providers has influenced the perception on how a cloud should provide its resources to end-users, i.e. on-demand and access to an unlimited amount of resources. However, not all clouds are equal. While such provisioning models work for well-endowed public clouds, they may not always work well in private clouds with limited budget and resources such as research and education clouds. Private clouds also stand to be impacted greatly by issues such as user resource hogging and the misuse of resources for nefarious activities. These problems are usually caused by challenges such as (1) limited physical servers/ budget, (2) growing number of users and (3) the inability to gracefully and automatically relinquish resources from inactive users. Currently, cloud resource management frameworks used for private cloud setups, such as OpenStack and CloudStack, only uses the pay-per-use model as the basis when provisioning resources to users. In this paper, we propose OpenStack Café, a novel methodology adopting the concepts of 'time' and booking systems' to manage resources of private clouds. By allowing users to book resources over specific time-slots, our proposed solution can efficiently and automatically help administrators manage users' access to resource, addressing the issue of resource hogging and gracefully relinquish resources back to the pool in resource-constrained private cloud setups. Work is currently in progress to adopt Café into OpenStack as a feature, and results of our prototype show promises. We also present some insights to lessons learnt during the design and implementation of our proposed methodology in this paper.
This is an author’s accepted version of an article published in the Proceedings of 2014 IEEE Seventh International Conference on Cloud Computing. © 2014 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.