Location-aware cloud native media service delivery

Abstract

This paper discusses the Facility for Large-scale Adaptive Media Experimentation (FLAME) Platform, a Servicebased Architecture that demonstrates the concept of cloud native service orchestration and routing for media applications. This enables automated provisioning and management of microservices delivering a service function chain, which affords considerable flexibility and control to achieve delivery of defined Quality of Service to users in the face of varying demand, while at reasonable cost. The architecture of the system is presented, together with an exemplar media application illustrating automated routing and dynamic, automated and location-based control of microservices. Automated event detection and policies control the scaling of localised edge services, and user requests are automatically routed to local services. Illustrative examples of the performance characteristics on a testbed platform comparing two scenarios - scaled-in (one central server) and scaled-out (local content serving at the edge) - are discussed, as well as the time to switch between the scaling scenarios based on sensed local demand. Results indicate that on the testbed platform, switching between scaled-in and scaled out takes in the order of 2-3 seconds, and the scaled-out scenario has between a four-fold and six-fold user-experience performance gain over the scaled-in scenario.

More information

Identifiers

ORCID for Steve Taylor: orcid.org/0000-0002-9937-1762

ORCID for Michael Boniface: orcid.org/0000-0002-9281-6095

ORCID for Stephen Phillips: orcid.org/0000-0002-7901-0839

Catalogue record

Export record