NeTS: Small: Strategic Management of Advance Reservations in Cloud and Network Services

预留(AR)服务构成经济的支柱。例如，它们被广泛部署在交通、住宿和医疗保健行业。它们也越来越多地被用于通信网络和云计算中的资源管理。对于用户，它们提供了资源在需要时可用的保证，而对于供应商，它们提供了新的收入来源和预测需求的能力。然而，支持AR的系统的行为是复杂的，因为客户关于预订的决定会影响其他客户的性能。此外，服务提供商采用的定价方案对客户行为有重大影响。该项目的主要目标是开发数学模型，用于描述用户和提供商在支持AR的云和网络系统中的战略行为。该项目还需要设计有效的资源分配和定价机制，以激励广泛采用AR服务，使客户和服务提供商都受益。第二部分本项目的主要目标是：i)开发模型来描述云计算和网络系统中采用的提前预订方案中所有参与者的策略行为；ii)使用博弈论分析这些模型的均衡结果；iii)设计有效的资源分配和定价机制，以激励所有参与者受益于AR服务的部署；以及iv)将理论转化为实践。该项目从多个维度融合了云和网络服务中提前预订的关键特征，包括需求的随机性、工作负载的异构性、网络流量和拓扑的动态性，以及用户根据历史数据学习和调整策略的能力。调查需要使用既有的和较新的概念来描述均衡结果及其效率，例如在非合作情况下的无政府状态的价格和保守主义的价格，以及在合作情况下的核心和讨价还价方案。该项目还包括对开放式云市场(马萨诸塞州开放式云)的测量，以便为分析提供信息并指导服务提供商和客户做出战略决策。

Part 1.Advance reservation (AR) services form a pillar of the economy. For instance, they are widely deployed in the industries of transportation, lodging, and health care. They are also increasingly being adopted for the management of resources in communication networks and in cloud computing. For users they offer assurance that resources will be available when needed, while for providers they offer a new stream of revenue and the ability to forecast demand. Yet, the behavior of systems supporting AR is complex, since the decision of a customer about making a reservation affects the performance of other customers. Furthermore, pricing schemes employed by service providers have a significant impact on customer behavior. The main goals of this project are to develop mathematical models for characterizing the strategic behavior of users and providers in cloud and networking systems that support AR. The project also entails the design of efficient resource allocation and pricing mechanisms to incentivize widespread adoption of AR services in a way that benefits both customers and the service provider. Part 2The main goals of this project are: i) to develop models for characterizing the strategic behavior of all participants in advance reservation schemes employed in cloud computing and network systems; ii) to analyze the equilibrium outcomes of these models using game theory; iii) to design efficient resource allocation and pricing mechanisms that incentivize the deployment of AR services to the benefit of all participants; and iv) to transition theory into practice. This project incorporates key features of advance reservation in cloud and network services along several dimensions, including the randomness of the demand, the heterogeneity of the workload, the dynamic nature of the network traffic and topology, and the ability for users to learn and adapt their strategies based on historical data. The investigations entail characterizing the equilibria outcomes and their efficiency, using both established and newer concepts, such as the Price of Anarchy and the Price of Conservatism in non-cooperative contexts, and the Core and Bargaining Schemes in cooperative contexts. The project further includes measurements on an open cloud marketplace (Massachusetts Open Cloud) to inform the analysis and guide both service providers and customers in their strategic decisions.