Modeling Accountable Cloud Services

Abstract

Cloud computing services have been increasingly considered by business as a viable option for reducing IT expenditure. The highly automated and agile nature of cloud services offer businesses low cost, high efficiency and flexibility benefits. However, there are often associated problems with unmanaged accountability such as lack of disclosure of service obligations, mechanisms for detection for obligation fulfilment or determination of liable party if an obligation is violated. This paper analyses the accountability properties of a cloud service and proposes an accountable cloud service (ACS) model to address those problems. The ACS model is underpinned by a hybrid logic system called Dynamic Logic for Accountability (DLA) extended from Dynamic Logic. ACS provides an intuitive notation for modeling service collaboration diagrams based on a reduced version of BPMN2.0 to capture the fulfillment of service obligations. We also propose an Obligation Flow Diagram (OFD) as a simple method for conflict resolution and verification for the ACS model. The ACS model enables obligation specification, decomposition, validation, machine-interpretation, monitoring and reasoning, and ultimately facilitates accountability in cloud service consumption. Using Amazon S3 service as a case study, we show how to address those known accountability problems using our ACS model. Finally we discuss the applicability of our model to cloud services in general.