Supporting placement and data consistency strategies using hybrid clouds

In this paper, we investigate cloud platform support that provides distributed applications with automatic service placement across different cloud computing systems (hybrid clouds), and that enables application developers to investigate the impact of using different cloud-based data consistency models with their applications. By pursuing its implementation at the cloud runtime layer (platform), we are able to provide hybrid cloud support without requiring application modification or significant developer expertise. We investigate the efficacy of such portabililty for different application domains (web services and computationally-intensive HPC applications, via Monte Carlo simulations). We examine different hybrid cloud placement strategies based on cost, performance, and common use cases. We evaluate the performance of such placement strategies as well as different data consistency policies.