This paper proposes a model to study the interaction of price competition and congestion in the cloud computing marketplace. Specifically, we propose a three-tier market model that captures a marketplace with users purchasing services from Software-as-a-Service (SaaS) providers, which in turn purchase computing resources from either Provider-as-a-Service (PaaS) or Infrastructure-as-a-Service (IaaS) providers. Within each level, we define and characterize market equilibria. Further, we use these characterizations to understand the relative profitability of SaaSs and PaaSs/IaaSs, and to understand the impact of price competition on the user experienced performance, i.e., the "price of anarchy" of the cloud marketplace. Our results highlight that both of these depend fundamentally on the degree to which congestion results from shared or dedicated resources in the cloud.
Many video-on-demand (VoD) providers are relying on public cloud providers for video storage, access and streaming services. In this paper, we investigate how a VoD provider may make optimal bandwidth reservations from a cloud service provider to guarantee the streaming performance while paying for the bandwidth, storage and transfer cost. We propose a predictive resource auto-scaling system that dynamically books the minimum amount of bandwidth resources from multiple servers in a cloud storage system, in order to allow the VoD provider to match its short-term demand projections. We exploit the anti-correlation between the demands of different videos for statistical multiplexing to hedge the risk of under-provisioning. The optimal load direction from video channels to cloud servers without replication constraints is derived with provable performance. We further study the joint load direction and sparse content placement problem that aims to reduce bandwidth reservation cost under sparse content replication requirements. We propose several algorithms, and especially an iterative L1-norm penalized optimization procedure to efficiently solve the problem while effectively limiting the video migration overhead. The proposed system is backed up by a demand predictor that forecasts the expectation, volatility and correlation of the streaming traffic associated with different videos based on statistical learning. Extensive simulations are conducted to evaluate our proposed algorithms, driven by the real-world workload traces collected from a commercial VoD system.
Ensuring predictability is a crucial goal for service systems. Traditionally, research has focused on designing systems that ensure predictable performance for service requests. Motivated by applications in cloud computing and electricity markets, this paper focuses on a different form of predictability: predictable allocations of service capacity. The focus of the paper is a new model where service capacity can be scaled dynamically and service deferrals (subject to deadline constraints) can be used to control the variability of the active service capacity. Four natural policies for the joint problem of scheduling and managing the active service capacity are considered. For each, the variability of service capacity and the likelihood of deadline misses are derived. Further, the paper illustrates how pricing can be used to provide incentives for jobs to reveal deadlines, and thus enable the possibility of service deferral in systems where the flexibility of jobs is not known to the system a priori.