Performance based Risk driven Trust (PRTrust): On modeling of secured service sharing in peer-to-peer federated cloud

Abstract

Peer-to-peer (P2P) federated cloud system has emerged as a promising service delivery model wherein multiple cloud service providers (CSPs) collaborate and share resources and services among themselves to fulfill the spikes in demand from their cloud service consumers (CSCs). It facilitates CSPs to accomplish the committed service level agreements (SLAs) to their CSCs. However, the lack of preexisting trust relationships and unawareness about the cloud infrastructure and service delivery performance among CSPs in a distributed environment poses a risk to the quality of service (QoS) being delivered. We address this challenge by proposing a trust model, PRTrust, for a peer-to-peer federated cloud system that capitalizes the triangular relationship of performance, risk, and trust for the participating CSPs. The main contributions of this work are: (1) providing a logical design and architecture for trust and performance management in a peer-to-peer federated cloud system; (2) providing a two-tier weighted performance evaluation mechanism for CSPs; (3) providing a risk evaluation mechanism for CSPs based on their current performance level; (4) providing an improvised mechanism to evaluate and manage personalized reputation-based trust for the participating CSPs; (5) providing a CSP selection mechanism from a trusted list of CSPs, using the evaluated performance-based risk, for sharing the resources and services in a peer-to-peer federated cloud system. The proposed PRTrust model has shown better threat resilient behavior in dealing with malicious peer CSPs when compared with the reference EigenTrust model.