Model-driven cloud resource management with OCCIware

Abstract

Abstract Cloud computing has emerged as the main paradigm for hosting and delivering computing resources as services over Internet. It provides a delivery model for computing resources at infrastructure, platform, and software levels. However, there is a plethora of cloud providers offering different resource management interfaces. Due to this diversity, the consumption, provisioning, management, and supervision of cloud resources are subjected to four key issues, i.e., heterogeneity, interoperability, integration, and portability. To cope with these issues, Open Cloud Computing Interface (OCCI) is proposed as a community-based and open recommendation standard for managing any kind of cloud resources. Currently, only runtime implementations exist for OCCI, and each one targets a specific cloud service model such as Infrastructure as a Service (IaaS), Platform as a Service (PaaS), or Software as a Service (SaaS). Thus, OCCI lacks an approach to model and execute different OCCI artifacts. Our approach provides a generic modeling framework coupled with a generic runtime implementation. In this article, we propose the OCCIware approach, which represents the first approach to design, validate, generate, implement, deploy, execute, and supervise everything as a service with OCCI. This approach provides OCCIware Studio , the first model-driven tool chain for OCCI. It is built around OCCIware Metamodel , which defines the static semantics for the OCCI standard in Ecore and OCL. In addition, it proposes OCCIware Runtime , the first generic OCCI runtime implementation targeting all the cloud service models (IaaS, PaaS, and SaaS). OCCIware provides a unique and unified framework to manage OCCI artifacts and, at the same time, it represents a factory to build cloud domain-specific modeling frameworks where each framework targets a specific cloud domain. OCCIware has been applied in various cloud domains and use-cases, which validate its applicability.