System and method for cloud-based live media ingestion and transcoding

摘要

Embodiments are provided for a cloud-based reliable media ingestion and transcoding. The embodiments comprise an improved backup mechanism where each worker node, such as a virtual machine, physical machine, workstation, or server, in a cloud or Internet environment handles one streamed media/video channel transcoding and another streamed media/video channel ingestion backup. To reduce ingestion storage demands, the ingestion backup storage is based on a pre-determined relatively short-term sliding window. When an ingestion and transcoding worker node fails, a new worker node is brought up to take over from the failed worker node the live transcoding task of a first channel and the ingestion backup task of a second channel. Additionally, the new worker node retrieves backup data for the first channel from a third worker node and a central media storage to ensure seamlessly uninterrupted live ingestion and transcoding of the first channel, and continues the ingestion backup task of the second channel.

主权项

1. A method for media ingestion and transcoding, the method comprising:
receiving, at an error recovery worker node, instruction to take over live ingestion and transcoding of a streamed channel assigned to an ingestion and transcoding worker node separate from the error recovery worker node; retrieving, at the error recovery worker node, stored data of the streamed channel, the stored data obtained from the ingestion and transcoding worker node; while retrieving the stored data of the streamed channel, performing live ingestion of real-time data of the streamed channel; storing the live ingested real-time data into a relatively short-term sliding window-based temporary storage at the error recovery worker node, wherein the relatively short-term sliding window is predetermined relative to content of the streamed channel; transcoding, at the error recovery worker node, the retrieved data for the streamed channel assigned to the ingestion and transcoding worker node until the transcoded data catches up to a live data point received at the error recovery worker node; and upon catching up to the live data point, resuming, at the error recovery worker node, live transcoding of data for the channel from the live data point.