Service Agreement Trifecta: Backup Resources, Price and Penalty in the Availability-Aware Cloud

• 成果类型: Article
• 署名作者: Yuan, Shuai; Das, Sanjukta; Ramesh, R.; Qiao, Chunming
• 署名单位: Brock University; State University of New York (SUNY) System; University at Buffalo, SUNY; State University of New York (SUNY) System; University at Buffalo, SUNY
• 刊物名称: INFORMATION SYSTEMS RESEARCH
• ISSN/ISSBN: 1047-7047
• DOI: 10.1287/isre.2017.0755
• 发表日期: 2018
• 页码: 947-964
• 关键词: risk-management; demand; INFORMATION; policies; DESIGN

摘要

Service Level Agreements (SLA) for cloud services entail complex trade-offs between interrelated variables such as price, penalty, and service availability (uptime) guarantee, with resource management strategies affecting fulfillment of the SLA. In this study, we address three key components of the SLA-based cloud resource management and pricing problem, from the service-provider's perspective: (1) availability-aware backup resource provisioning; (2) price-penalty schedule determination; and (3) penalty-deferred pricing over two periods. Using the convexity of the provider's expected total cost over the number of backup resources, we present a dichotomous search algorithm to derive the total cost minimizing number of backup resources for a given level of SLA-specified service availability guarantee. Next, we derive closed-form solutions for the lower bound of the feasible price range, yielding a schedule of breakeven price-penalty combinations, which establishes the baseline required in the economic modeling of the service contracts and related negotiation processes, and may also elicit client preference information. We then model a two-period pricing problem specifically designed to incentivize penalty deferrals in the event of an SLA violation. Detailed experimental studies of the proposed models have been carried out using real-world datacenter log data. The computational study validates the convexity of the probability density function of SLA violations over the number of backup resources. The results demonstrate significant interaction effects between the SLA parameters (price, penalty rate, and provisioning cost) and the backup resource provisioning decisions made by the provider, leading to key practical managerial implications for SLA design and resource deployment in the availability-aware cloud.