Description

Abstract

          Due to dynamic resource provisioning technologies used in cloud, the frequent resource repartitioning and reallocation (e.g., upon task arrival or completion) make it a challenging problem to locate a node containing a combination of available resources along all the R resource attributes that would satisfy the requirements of a submitted task. In recent years, various enhancements on resource isolation techniques have been proposed to achieve fine-grained dynamic resource provisioning. we propose a novel cloud architecture, namely self-organizing cloud (SOC), which can connect a large number of desktop computers on the Internet by a P2P network. In SOC, each participating computer acts as both a resource provider and a resource consumer. They operate autonomously for locating nodes with more abundant resource or unique services in the network to offload some of their tasks, meanwhile they could construct multiple VM instances for executing tasks submitted from others whenever they have idle resources. we propose an algorithm to optimize the task execution time on a qualified resource node, given its preset budget and tolerable quality of service (QoS).