Video streams usually have to be transcoded to match the characteristics ofviewers' devices. Streaming providers have to store numerous transcodedversions of a given video to serve various display devices. Given the fact thatviewers' access pattern to video streams follows a long tail distribution, forthe video streams with low access rate, we propose to transcode them in anon-demand manner using cloud computing services. The challenge in utilizingcloud services for on-demand video transcoding is to maintain a robust QoS forviewers and cost-efficiency for streaming service providers. To address thischallenge, we present the Cloud-based Video Streaming Services (CVS2)architecture. It includes a QoS-aware scheduling that maps transcoding tasks tothe VMs by considering the affinity of the transcoding tasks with the allocatedheterogeneous VMs. To maintain robustness in the presence of varying streamingrequests, the architecture includes a cost-efficient VM Provisioner. Thiscomponent provides a self- configurable cluster of heterogeneous VMs. Thecluster is reconfigured dynamically to maintain the maximum affinity with thearriving workload. Results obtained under diverse workload conditionsdemonstrate that CVS2 architecture can maintain a robust QoS for viewers whilereducing the incurred cost of the streaming service provider up to 85%
展开▼
