The performance of large-scale peer-to-peer (P2P) video-on-demand (VoD) streaming systems can be very challenging to analyze due to sparse connectivity and complex, random dynamics. Specifically, in practical P2P VoD systems, each peer only interacts with a small number of other peers/neighbors. Furthermore, its upload capacity, downloading position, and content availability change dynamically and randomly. In this paper, we rigorously study large-scale P2P VoD systems with sparse connectivity among peers and investigate simple and decentralized P2P control strategies that can provably achieve close-to-optimal streaming capacity. We first focus on a single streaming channel. Using a simple algorithm that assigns each peer a random set of Θ(logN) neighbors and allocates upload capacity uniformly, we show that a close-to-optimal streaming rate can be asymptotically achieved for all peers with high probability as the number of peers N increases. Furthermore, the tracker does not need to obtain detailed knowledge of which chunks each peer caches, and hence incurs low overhead. We then study multiple streaming channels where peers watching one channel may help peers in another channel with insufficient upload bandwidth. We propose a simple random cache-placement strategy and show that a close-to-optimal streaming capacity region for all channels can be attained with high probability, again with only Θ(logN) per-peer neighbors. These results provide important insights into the dynamics of large-scale P2P VoD systems, which will be useful for guiding the design of improved P2P control protocols.

Latest developments in cloud computing technologies have enabled a plethora of cloud based data storage services. Cloud storage service providers are facing significant bandwidth cost as the user population scales. Such bandwidth cost can be substantially slashed by exploring a hybrid cloud storage architecture that takes advantage of under-utilized storage and network resources at storage clients. A critical component in the new hybrid cloud storage architecture is an economic mechanism that incentivizes clients to contribute their local resources, while at the same time minimizes the provider's cost for pooling those resources. This work studies online procurement auction mechanisms towards these goals. The online nature of the auction is in line with asynchronous user request arrivals in practice. After carefully characterizing truthfulness conditions under the online procurement auction paradigm, we prove that truthfulness can be guaranteed by a price-based allocation rule and payment rule. Our truthfulness characterization actually converts the mechanism design problem into an online algorithm design problem, with a marginal pricing function for resources as variables set by cloud storage service providers for online procurement auction. We derive the marginal pricing function for the online algorithm. We also prove the competitive ratio of the social cost of our algorithm against that of the offline VCG mechanism and of the resource pooling cost of our algorithm against that of the offline optimal auction. Simulation studies driven by real-world traces are conducted to show the efficacy of our online auction mechanism.

Cloud providers often choose to operate datacenters over a large geographic span, in order that users may be served by resources in their proximity. Due to time and spatial diversities in utility prices and operational costs, different datacenters typically have disparate charges for the same services. Cloud users are free to choose the datacenters to run their jobs, based on a joint consideration of monetary charges and quality of service. A fundamental problem with significant economic implications is how the cloud should price its datacenter resources at different locations, such that its overall profit is maximized. The challenge escalates when dynamic resource pricing is allowed and long-term profit maximization is pursued. We design an efficient online algorithm for dynamic pricing of VM resources across datacenters in a geo-distributed cloud, together with job scheduling and server provisioning in each datacenter, to maximize the profit of the cloud provider over a long run. Theoretical analysis shows that our algorithm can schedule jobs within their respective deadlines, while achieving a time-average overall profit closely approaching the offline maximum, which is computed by assuming that perfect information on future job arrivals are freely available. Empirical studies further verify the efficacy of our online profit maximizing algorithm.

The performance of large-scaled peer-to-peer (P2P) video-on-demand (VoD) streaming systems can be very challenging to analyze. In practical P2P VoD systems, each peer only interacts with a small number of other peers/neighbors. Further, its upload capacity may vary randomly, and both its downloading position and content availability change dynamically. In this paper, we rigorously study the achievable streaming capacity of large-scale P2P VoD systems with sparse connectivity among peers, and investigate simple and decentralized P2P control strategies that can provably achieve close-to-optimal streaming capacity. We first focus on a single streaming channel. We show that a close-to-optimal streaming rate can be asymptotically achieved for all peers with high probability as the number of peers N increases, by assigning each peer a random set of Θ(log N) neighbors and using a uniform rate-allocation algorithm. Further, the tracker does not need to obtain detailed knowledge of which chunks each peer caches, and hence incurs low overhead. We then study multiple streaming channels where peers watching one channel may help in another channel with insufficient upload bandwidth. We propose a simple random cache-placement strategy, and show that a close-to-optimal streaming capacity region for all channels can be attained with high probability, again with only Θ(log N) per-peer neighbors. These results provide important insights into the dynamics of large-scale P2P VoD systems, which will be useful for guiding the design of improved P2P control protocols.