Audit meets game theory: Verifying reliable execution of SLA for compute-intensive program in cloud

Cloud computing provides convenient on-demand computing resources which enables users to outsource their computational tasks to the cloud. However, users lose direct control of tasks, which inevitably causes new challenges: users have no way to audit whether the cloud server provides services along the service level agreement (SLA). To this end, we present an audit model based on cloud service broker (CSB) that can verify the reliable execution of SLA for compute-intensive program in cloud. Specifically, we propose a program structure mapping model called concept tree, which transforms the target application into a tree structure. Then, we split it into several independent computable units. Based on this, CSB can fine-grained audit the task execution process in cloud by comparing the running time and the similarity of the computable units. We also adopt game theory to motivate the credible implementation of SLA, in addition to defend against the semi-honest cloud model. We implement our scheme on Hadoop and the extensive experiments demonstrate that our scheme has the high prediction accuracy.