Utility-Driven Adaptive Scheduling for Cloud Service Provisioning
Manisha T. Tapale1, R. H. Goudar2, Mahantesh N. Birje3
1Manisha T. Tapale, Department of Computer Science and Engineering, KLE Dr. MSSCET, Belagavi, India.
2R. H. Goudar, Center for Post Graduation Studies, Visvesvaraya Technological University, Belagavi, India.
3Mahantesh N. Birje, Center for Post Graduation Studies, Visvesvaraya Technological University, Belagavi, India.
Manuscript received on 02 June 2019 | Revised Manuscript received on 10 June 2019 | Manuscript published on 30 June 2019 | PP: 249-257 | Volume-8 Issue-8, June 2019 | Retrieval Number: H6327068819/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Though multiple cloud service providers (CSPs) offer similar services the access patterns of these services vary in general. Also, the CSPs belonging to different administrative domains will have different policies, preferences, and objectives for increasing their business profit as well as resource utilization. While Cloud user expects its service requests to be served as early as possible, CSPs try to serve user requests within deadline and obtain more profit (utility). Such utility motivated service provisioning may affect low utility jobs to starve from resources. Hence it is very much essential for CSPs to consider a deadline of jobs to avoid starvation. To address this issue we propose a utility-driven adaptive scheduling scheme that works as follows: 1) The service price, and thus the utility of service provisioning is determined using Non-cooperative bargaining protocol, 2) The scheduler prioritizes jobs depending on their utility and deadline, and then orders them in priority queue for execution. 3) On peak loads as new jobs with high utility arrive, some of the existing jobs having lesser utility in a priority queue start suffering from starvation. In such context, the proposed scheme adapts to generate another queue, called as FCFS queue, which holds relatively lesser profit jobs from priority queue. 4) The dispatcher then dispatches jobs to reliable virtual machines, choosing alternatively from priority queue and FCFS queue to achieve fair scheduling. The proposed scheme aims at maximization of the profit gained by CSP and avoidance of job starvation along with minimization of the service completion time and maximization of the utilization of resources. The proposed work is simulated using CloudSim and results show that it performs better than existing works.
Keyword: Adaptive Scheduling, Cloud Computing, Priority, , Starvation, Service Provisioning, Utility.
Scope of the Article: Mobile Cloud Computing and Application Services