Title :
A Method for Solving the Performance Isolation Problem in PaaS Based on Forecast and Dynamic Programming
Author :
Yu, Jiayang ; Rao, Ruonan
Author_Institution :
Sch. of Software, Shanghai JiaoTong Univ., Shanghai, China
Abstract :
As a Platform as a Service (PaaS) designer, we should satisfy customers´ demands of resources and to save resources in a cloud at the same time. In PaaS, one of the main problems is that if one of virtual machines´ (VM) workload increases fast in a host, other VMs´ performance characteristics may be influenced negatively. To solve the performance isolation problem, this paper gives a method. In this method, we use an algorithm to forecast all VMs´ resources usage at first. Then we find which VMs´ workloads may increase fast in future. We move these VMs to hosts which have more resources and other VMs´ performance characteristics will not be affected at the same time. So the main problem now is which hosts we should use to move these VMs as mentioned above to. In this paper, we use Dynamic Programming method to solve this problem. So, with all of these algorithms, in this paper, we propose a strategy to solve Performance Isolation problem in PaaS.
Keywords :
cloud computing; dynamic programming; resource allocation; virtual machines; PaaS; VM resource forecasting; VM workload; cloud computing; dynamic programming; performance isolation problem; platform as a service; virtual machine workload; Cloud computing; Curve fitting; Dynamic programming; Educational institutions; Monitoring; Resource management; Storage area networks; PaaS; cloud computing; dynamic programming; forecast; performance isolation;
Conference_Titel :
Computational and Information Sciences (ICCIS), 2012 Fourth International Conference on
Conference_Location :
Chongqing
Print_ISBN :
978-1-4673-2406-9
DOI :
10.1109/ICCIS.2012.23