Efficient allocation of elastic interfederation encrypted streaming sessions

Abstract

SummaryAudio and video (A/V) collaboration platforms often use Internet cloud technologies to ensure elasticity. They generally operate on a best‐effort basis, without quality or delivery guarantees. However, such guarantees are a premise of business‐focused platforms, which often rely on static/dedicated infrastructure and hardware‐based components. This article presents results obtained in the final stage of the Elastic Media Distribution (EMD) project, which targets the migration of a business‐focused hardware‐based A/V collaboration tool towards a more elastic, reliable, and secure cloud‐based model. The use case under investigation is an educational scenario: teachers and students located at distributed sites collaborate under different data encryption policies. An existing model of collaboration streaming is extended to accommodate encryption‐enabled streaming components, and new resource allocation heuristics are proposed to deploy these components under stringent service level agreement (SLA) constraints. An extended version of our evaluation framework, based on the CloudSim simulator, manages encryption‐enabled components. A resource usage dataset was obtained by prototyping selected streaming components and evaluating their performance on the Virtual Wall large‐scale test bed. This dataset is fed into the extended simulation framework. Simulation results show longer than expected delays when loading streaming components, an issue that jeopardises the user experience that can be alleviated by the algorithms proposed in this article. Results show that the proposed algorithms enable policy‐based secured communications under bandwidth and virtual machine (VM) cost increases of 48% and 23%, respectively, if compared with a nonencrypted previous solution, and with set‐up times remaining under the required 2‐second deadline.