A Cloud Federation Architecture

Abstract

Cloud Computing is the state of the art computation paradigm enabling providers to offer computing resources to customers in a pay-per-use fashion. Nowadays public Cloud providers offer similar services within few service models, mainly IaaS and PaaS. Cloud providers give to the user the feeling to dispose of infinite resources, thus having to predict the user requirements in order to provide services with minimal costs, maintaining at the same time high levels of SLAs. In order to achieve this goal, Cloud providers can cooperate together to bring new business opportunities, such as expanding available resources, achieving cost effective asset optimization and adopting power saving policies. Cloud Federation allows different Cloud providers the opportunity to work collaboratively to offer best services to customers and contemporary to improve their productivity. Customers can advantage from Cloud Federation for a larger offer of available services, the capability of price comparison and the removal of vendor lock-in. In this paper we describe a platform that enables the federation of several heterogeneous Cloud Providers to allow the customers choosing and activating Cloud services from a central platform, bringing more attractive price policy to customers. The authors introduce a prototype of Cloud Federation platform based on a central infrastructure tested to manage OpenStack, CloudStack and Amazon EC2 providers, thus allowing the user to select the best services in terms of either technical requirements or price policy and activate them without having to explicitly register to each of the federated providers. The prototype is designed to accept different types of Cloud service models by means of transparent interfaces developed around a billing and a metering module, respectively to bill the service to the customer and to collect information about the health status of the federated platforms.