Dynamic Resource Allocation for Cloud-based （动态资源分配的云）.pdf

Dynamic Resource Allocation for Cloud-based Media Processing Krisantus Sembiring Andreas Beyer NEC Labs Europe Ltd. NEC Labs Europe Ltd. Kurfürsten-Anlage 36 Kurfürsten-Anlage 36 69115 Heidelberg, Germany 69115 Heidelberg, Germany krisantus.sembiring@neclab.eu andreas.beyer@neclab.eu ABSTRACT alization technologies and the pay-as-you-use model. As an economic and scalable solution of providing interactive and adaptive media content across different devices, cloud- Within a cloud-based media processing platform, a large based media processing has recently attracted lots of at- number of media processing pipelines will be spawn to run in tention from both academic and industry. Within a media parallel across virtural machines and phyisical hosts. Each cloud, a large number of media processing tasks will be dy- task takes content in, processes the content, and offers it namically spawn to run in parallel and then share resource to others. They are highly heterogeneous and dynamic in with each other in the cloud. In such a dynamic and shared terms of resource consumption. For example, a pipeline gen- environment, how to appropriately provision cloud resource erating WEBM stream from MPEG2 stream will consume to different tasks with regards to both system efficiency and much more CPU than transcoding mp3