Instant social video sharing which combines the online social network anduser-generated short video streaming services, has become popular in today'sInternet. Cloud-based hosting of such instant social video contents has becomea norm to serve the increasing users with user-generated contents. Afundamental problem of cloud-based social video sharing service is that usersare located globally, who cannot be served with good service quality with asingle cloud provider. In this paper, we investigate the feasibility ofdispersing instant social video contents to multiple cloud providers. Thechallenge is that inter-cloud social \emph{propagation} is indispensable withsuch multi-cloud social video hosting, yet such inter-cloud traffic incurssubstantial operational cost. We analyze and formulate the multi-cloud hostingof an instant social video system as an optimization problem. We conductlarge-scale measurement studies to show the characteristics of instant socialvideo deployment, and demonstrate the trade-off between satisfying users withtheir ideal cloud providers, and reducing the inter-cloud data propagation. Ourmeasurement insights of the social propagation allow us to propose a heuristicalgorithm with acceptable complexity to solve the optimization problem, bypartitioning a propagation-weighted social graph in two phases: apreference-aware initial cloud provider selection and a propagation-awarere-hosting. Our simulation experiments driven by real-world social networktraces show the superiority of our design.
展开▼