• UID623
  • Fans2
  • Follows1
  • Posts72

[Reference architecture] On-demand video streaming

More Posted time:Dec 5, 2016 13:43 PM
Industry requirements
On-demand video streaming has become a quite common business scenario in the internet sector. It has two primary business models: one is represented by video websites, which scenario features little uploading but massive demands for video resources; the other is home surveillance video demand, which scenario features massive uploading but little demand for video resources. The on-demand video streaming business faces the following problems in the traditional IT environment:
1. High resource consumption and fast increase. Unlike Web services, the on-demand video streaming business is highly storage resource-consuming. A sizable on-demand video streaming website may have several hundred TBs or PBs of stored resources and the limited infrastructure of general IDCs or small cloud service providers is easy to become a bottleneck constraining the explosive growth of on-cloud video streaming business. In addition, the traditional IT environment is hard to be scaled with a high migration cost.
2. Sensitive to network bandwidth and network quality. HD and being smooth are the most important user experiences for the on-demand video streaming business. HD videos have a high bit rate and require sufficient and high-quality network bandwidth to ensure latency during playback is within a tolerable range and guarantee smooth playing amid a high concurrency access scenario. What's more, the high-bandwidth costs arising from video distribution should also be considered.
3. The content is the core value and must be effectively protected against leeches.

The following solution is designed to respond well to the above challenges. At the same time, it can effectively reduce IT costs.

About the solution
1. Videos are recorded by video capture devices or services and uploaded to the cluster for receiving video files. The video file-receiving cluster has multiple Web servers mounted to the SLBs. This structure enables the cluster to be horizontally scalable. Considering the peak fluctuations of the video uploading business, you can adopt the Elastic Scaling Service (ESS) to achieve automatic adjustment of the ECS amount in the cluster dynamically to adapt to the actual business peaks.
2. After a video file-receiving server receives the file, it stores the spliced file into the Open Storage Service (OSS). OSS is capable of massive storage and can serve as the storage repository of massive tutorial videos.
3. The video file-receiving server then commits a transcoding operation to the transcoding platform to convert the tutorial videos into multiple video formats for various clients. Alibaba Cloud's transcoding platform supports rich input file formats and mainstream output file formats, such as FLV, MP4, and M3U8 TS splices.
4. The video website service is a Web service providing browsing of online tutorials and user management. If the website has a large user base and a considerable page view, the SLB can also be considered to achieve highly available and scalable clusters.
5. The streaming media forwarding server provides the on-demand teaching video service to authorized users. With the combination of SLB and ESS, the cluster can be horizontally expanded and dynamically scalable to cope with the fluctuated business visits.
6. This solution adopts CDN to accelerate video streaming so that users in various regions can enjoy a more smooth watching experience.

Features and advantages
1. Elastic computing efficiently reduces the computing cost. Cost-efficient and low-configuration Elastic Cloud Service (ECS) is used to ensure service continuity. Elastic Scaling Service (ESS) is used to automatically increase the cloud server quantity in the service peak hours and thus ensure proper functioning of the service. This approach can effectively cut down the computing cost and streamline the overall cost structure.
2. Better video watching experience. The superior BGP bandwidth of Alibaba Cloud guarantees lower playback latency: Alibaba Cloud covers more than 100 CDN nodes distributed nationwide to provide the acceleration service for videos on demand, bringing a faster and smoother video watching experience.
3. Massive video file storage. Alibaba Cloud's OSS provides the massive file storage service and features data reliability up to 99.99999999%. It also provides local hot backup for disaster recovery and remote disaster recovery solutions to ensure permanent storage and availability of massive monitoring videos.
4. Enhanced scalability. In this solution, the storage, computing, and distribution schemes all allow elastic resizing from basic configurations to tens of thousands cores, dozens PBs and several Tbps of scale. In addition, Alibaba Cloud's scale advantages ensure sufficient inventory.
5. Complexity is reduced in implementing the transcoding service. Alibaba Cloud's transcoding platform provides a multiline and efficient transcoding service and supports multiple input/output formats. You don't need to implement transcoding algorithms on your own, reducing the business complexity and computing costs.