| Cloud computing has emerged as the mainstay of cost-effective, high-performance platforms for hosting personal and organizational information assets and computing tasks. This thriving computing paradigm, however, also comes with security perils. Recent years have seen a rising trend of security breaches in the cloud. In this dissertation, we will propose novel techniques to understand and solve some of the security and privacy challenges of cloud computing.;First we present the design and implementation of a new framework to support high-performance auto-scaling detection services for cloud. This framework has been fully integrated into the existing cloud orchestration mechanism, allowing an ordinary cloud user to request a detection service and specify its parameters conveniently, through the cloud-formation file she submits for deploying her application.;Then we report a suite of new techniques that achieve secure and scalable read mapping on hybrid clouds. Our approach leverages the special features of the read mapping task, which only cares about small edit distances, and of the Cloud, which is good at handling a large amount of simple computation. These features enable us to split the mapping computation according to the seed-and-extend strategy.;Finally we examine security issues of mobile push messaging, which is one of the most important and popular mobile-cloud services. We will present the design and implementation of an automatic analysis tool for identifying vulnerabilities in push messaging SDKs and mobile applications that leverage these SDKs. The study reveals the serious security challenges those mobile cloud services are facing and the urgent need for the efforts to ensure their security qualities. |
