Enhancing Real-Time Applications by Means of Multi-tier Cloud Federations
Vamis Xhagjika, Leandro Navarro, Vladimir Vlassov · 2015
The evolution of Cloud Computing beyond the frontier of a single datacenter is justified as a mean to enhance various system architecture aspects like: cost, geo-locality, energy efficiency and structural properties. Orchestration of different cloud providers in order to ensure better Quality of Service (QoS) is refered to as Multi-Cloud. One such multi-cloud federation model is Community Clouds, a federation concept in which various micro-clouds are owned by different communities of individuals contributing cloud resources. Community Clouds provide high locality to services towards the user by bringing the Cloud in the edges of the network, and in some cases as an integral part of the access network. Another such type of multicloud federation is Structured Cloud Federation for Internet Service Providers and Telephony Providers, composed of geo distributed micro-clouds placed on the edges of the network and a centralized datacenter cloud on the core network. Services deployed in the micro-clouds have augmented locality, while services on the datacenter cloud enhanced performance. In this work we identify how these approaches to multi-cloud, through the ability to deploy services close to the final user, can be exploited to enhance existing real-time streaming applications. Furthermore a novel multi-cloud overlay algorithm is introduced that provides both latency and backbone traffic optimization for latency critical existing and novel services (concretely focusing on stream computation, live streaming).