| Cloud computing is a computing approach that refers to the on-demand provision of computational resources via the Internet or in private via the internal enterprise network, with a pay-as-you-use billing model. The paradigm IaaS (Infrastructure as a Service) provides users with an on-demand self-service access, to a virtualized IT infrastructure, often composed of virtual machines on which users can install, manage and customize their applications. Alternatively, PaaS (Platform as a Service) offers users a programming environment immediately available and fully manageable to create and deploy scalable applications in the cloud without user intervention. Although the choice of such an environment may seem advantageous enough, several challenges remain to effectively use cloud computing systems. Thus, cloud computing services are offered at different levels of abstraction where the cloud providers expose access to their services via proprietary APIs. This encourages vendor lock-in and limits the interoperability of cloud computing services, which constitutes a significant barrier to entry for cloud users. Several solutions have been proposed based on the use of intermediate layers that isolate applications from the variability of services offered by cloud providers. However, these approaches provide a partial solution to this problem when such solutions use certain proprietary technologies as they contribute to the risk of moving the effect of vendor lock-in from cloud providers to deployment tools. The main objective of our research is to design and develop a new generic model for automated deployment of applications in the cloud to mitigate the effects of these barriers to entry, reduce the complexity of application development and simplify the process of deploying cloud services. In addition, support and automatically deploy applications on the cloud by providing elasticity, making automatic-scaling and interoperability with all platforms and optimizing storage management are the primary objectives of this memory. Our proposed model, OpenICRA, implements a layered architecture that hides the implementation details, allowing having a simple deployment process. Furthermore, unlike other cloud computing solutions available such as Google App Engine, Windows Azure or Amazon Elastic Beanstalk, the components of the proposed model is characterized by the freedom to access to their source code. This allows us to ensure application portability in any execution environment, avoid vendor lock-in and facilitate automation process of applications in the cloud. The methods of redundancy, scaling and integration of distributed file system with the IaaS layer are used to provide high availability, scalability and extensibility of the model and applications as well as optimizing storage management of VMs virtual hard disks in any cloud environment. We conducted two real case studies to validate our proposed model, where the first consists to automate the deployment of a distributed middleware called OpenSAF across a cluster of nodes in the Cloud environment of GSN Network (Synchromedia, 2010), while the second case is to ensure the migration of the collaborative work application, ICRA (Cheriet, 2012) to Amazon EC2 (Amazon, 2012a). Our empirical results demonstrate the effectiveness of our proposed model to deploy different types of applications without any change in their source code. In addition, we show how our proposal is able to automate and orchestrate the process of application deployment and optimize their implementation in terms of performance in heterogeneous cloud environments. |
