Cloud execution environment for real-time media applications

Abstract: Smartphones and other Internet of Things devices have become a rapidly growing topic. Along with the growth comes new technologies, likeWeb Real- Time Communication (WebRTC), that enables richer services to be built for the devices. These kind of services are typically consumed on-demand, in shorter periods at a time. Likewise have cloud computing exploded in popularity during the last years. Cloud computing offers compelling advantages, such as rapid elasticity and on-demand usage, that allow servers' resource utilization to be more effcient. The flexibility of allocating and releasing resources swiftly as they are required, enables services that run in the cloud to adopt to ephemeral workloads. The research in this thesis targets a real-time video streaming service that is based on WebRTC. Incoming streams are handled by Multipoint Control Units (MCUs) that have the responsibility to redistribute the incoming streams to the consumers. Scaling horizontally aligns well with the idea of cloud computing. The work in this thesis is based on the extreme case where each of the incoming streams are handled by a separate MCU. The thesis presents the process of finding a exible Cloud Execution Environment (CEE) for the streaming service. The process includes an analysis of the streaming service's requirements, an evaluation of existing solutions, and an implementation. Moreover, the thesis includes a discussion about the capabilities of the implemented system. The result of the thesis is a CEE upon which the streaming service can be deployed and managed. The developed CEE allows any workload that is encapsulated within a Docker container to be orchestrated, not exclusively the streaming service, which makes the implementation viable to other cloud computing projects.