Search

The cloud computing services have gained popularity in both public and enterprise domains and they process a large amount of user data with varying privacy levels. The increasing demand for cloud services including storage and computation requires new functional elements and provisioning schemes to meet user requirements. Multi-clouds can optimise the user requirements by allowing them to choose best services from a large number of services offered by various cloud providers as they are massively scalable, can be dynamically configured, and delivered on demand with large-scale infrastructure resources. A major concern related to multi-cloud adoption is the lack of models for them and their associated security issues which become more unpredictable in a multi-cloud environment. Moreover, in order to trust the services in a foreign cloud users depend on their assurances given by the cloud provider but cloud providers give very limited evidence or accountability to users which offers them the ability to hide some behaviour of the service. In this thesis, we propose a model for multi-cloud collaboration that can securely establish dynamic collaboration between heterogeneous clouds using the cloud on-demand model in a secure way. Initially, threat modelling for cloud services has been done that leads to the identification of various threats to service interfaces along with the possible attackers and the mechanisms to exploit those threats. Based on these threats the cloud provider can apply suitable mechanisms to protect services and user data from these threats. In the next phase, we present a lightweight and novel authentication mechanism which provides a single sign-on (SSO) to users for authentication at runtime between multi-clouds before granting them service access and it is formally verified. Next, we provide a service scheduling mechanism to select the best services from multiple cloud providers that closely match user quality of service requirements (QoS). The scheduling mechanism achieves high accuracy by providing distance correlation weighting mechanism among a large number of services QoS parameters. In the next stage, novel service level agreement (SLA) management mechanisms are proposed to ensure secure service execution in the foreign cloud. The usage of SLA mechanisms ensures that user QoS parameters including the functional (CPU, RAM, memory etc.) and non-functional requirements (bandwidth, latency, availability, reliability etc.) of users for a particular service are negotiated before secure collaboration between multi-clouds is setup. The multi-cloud handling user requests will be responsible to enforce mechanisms that fulfil the QoS requirements agreed in the SLA. While the monitoring phase in SLA involves monitoring the service execution in the foreign cloud to check its compliance with the SLA and report it back to the user. Finally, we present the use cases of applying the proposed model in scenarios such as Internet of Things (IoT) and E-Healthcare in multi-clouds. Moreover, the designed protocols are empirically implemented on two different clouds including OpenStack and Amazon AWS. Experiments indicate that the proposed model is scalable, authentication protocols result only in a limited overhead compared to standard authentication protocols, service scheduling achieves high efficiency and any SLA violations by a cloud provider can be recorded and reported back to the user.