CAREER: From Storm Clouds to EverClouds: Heading Off Long-Term Cloud Computing Risks

Yale University, New Haven CT

Investigators

Abstract

The cloud computing model, however appealing from economic and management perspectives, brings risks that should be understood before too much critical infrastructure "moves to the cloud." While the privacy and information security risks are well-known, cloud computing may bring other long-term risks that are neither widely recognized nor addressed. EverCloud is a novel cloud computing model that explores and attempts to mitigate these risks. First, clouds exacerbate security risks from timing side-channels, which current approaches cannot address without undermining the cloud business model. EverCloud introduces a Timing Information Flow Control (TIFC) model, providing a tool to reason about and limit timing channels despite fine-grained hardware resource multiplexing. Second, cloud services built atop one another risk hidden interactions that might lead to unpredictable instabilities or "meltdowns," as well as resource interdependencies that undermine a service's resilience to failure. EverCloud adapts labeling techniques to enforce stability constraints on reactive cloud computations, and to ensure independence in the provisioning of cloud hardware infrastructure. Third, clouds exacerbate digital preservation challenges, because no one but a cloud application's provider can archive a "working" copy of the application and its data. EverCloud enables any stakeholder (e.g., customer, librarian) to create and independently preserve fully-functional "snapshots" of cloud applications, while mitigating the computing, storage, and bandwidth costs of preservation via proportionate cost-sharing among stakeholders. By exploring these possible cloud computing risks and potential solutions, EverCloud may increase the security, robustness, and predictability of cloud computing systems throughout the many areas in which cloud computing is increasingly used.