On the Load Balancing of Edge Computing Resources for On-Line Video Delivery

Online video broadcasting platforms are distributed, complex, cloud oriented, scalable, micro-service-based systems that are intended to provide over-the-top and live content to audience in scattered geographic locations. Due to the nature of cloud VM hosting costs, the subscribers are usually served under limited resources in order to minimize delivery budget. However, operations including transcoding require high-computational capacity and any disturbance in supplying requested demand might result in quality of experience (QoE) deterioration. For any online delivery deployment, understanding user’s QoE plays a crucial role for rebalancing cloud resources. In this paper, a methodology for estimating QoE is provided for a scalable cloud-based online video platform. The model will provide an adeptness guideline regarding limited cloud resources and relate computational capacity, memory, transcoding and throughput capability, and finally latency competence of the cloud service to QoE. Scalability and efficiency of the system are optimized through reckoning sufficient number of VMs and containers to satisfy the user requests even on peak demand durations with minimum number of VMs. Both horizontal and vertical scaling strategies (including VM migration) are modeled to cover up availability and reliability of intermediate and edge content delivery network cache nodes.