First, the program background
HD booming ultra high-definition video, users desire for high-quality video experience for concurrent processing and storage capacity of the network content platform put forward higher requirements. As an important part of the industrial chain, CDN (content delivery network) specification into the fast lane. The CDN content distribution platform for the rapid development of the construction to provide strong support and popularity. With the continuous development of business, with scale, resilience adaptive, client support massive concurrent access to storage platforms conform to the trend of the era of ultra-high-definition video.
Second, the challenge
1, the traditional source station storage can not meet new business needs
Source station video on demand platform is the core of the library, over time, the procurement of program content and more and more content storage needs the support of dozens or even hundreds of PB, is limited to the nose or architecture, traditional storage platform and can not store all content, and can not be achieved in the face of non-stop expansion of capacity is needed.
2, there are security problems of traditional data protection methods
Due to capacity limitations and other architectural issues, often using multiple sets of traditional storage memory for data storage, and each store can only use RAID and other mechanisms to ensure data security. As the amount of data increases, more and more storage nodes, a single point of failure issues highlighted.
3, traditional storage can not cope with the huge pressure of the mass flow rate of the client
Front-end user growth brings higher flow rate output, but the traditional mode backend storage is limited to the nose, stable output video stream can not continue to enhance the storage through expansion, in order to improve the user viewing experience, CDN and need other lines, etc. link video streaming accelerated processing.
4, traditional storage can not support the demand for fast indexing and video processing
Traditional storage does not support indexing the content, not by index, cache, stream splitting, and multicast technology, content distribution or delivered to the service node closest to the user directly, can not support video analysis, content review, smart intelligent video production processing engine.
Third, the solution
CDN video storage solutions provide a unified storage platform source station video data. By Shan rock mass object storage (SandStone MOS) local disks on multiple server consolidation to form one or more large-scale storage resource pool through multiple copies and data security protection mechanism failed domain, ensure data security and business continuity sex.
CDN is less write to read application scenario, when data is written, the stored data slice uniformly redundant distributed over different servers of different hard disk; when reading data, the plurality of nodes are simultaneously read from a hard disk, provide stable high bandwidth video stream, in order to meet the needs of front-end and more complicated, but also comes with storage load balancing service does not appear Caton network bottlenecks and ensure high-quality video experience.
Fourth, the program returns
l distributed data storage, high reliability elastic expansion
The use of decentralized, distributed architecture, the storage capacity of up to and above EB
On-demand procurement, capacity and performance increases linearly with the number of nodes increases
A variety of security mechanisms to protect multiple copies of data, such as multi-domain security failure
When hardware exceptions, quick start automatic data reconstruction mechanism
l video optimization technology, faster access to video experience
Slice large file data storage, providing multi-node, high-bandwidth video stream stability
Storage node online expansion, performance of the output increases linearly, the underlying data is automatically balanced distribution
Video traffic load balancing, the front end of the support concurrent reads mass content
l video tag indexing and intelligent video processing
Built-in video content analysis, content review, intelligent production of intelligent video processing engine
Automatic Video label processing, and more quickly, accurately retrieve video
Support and Hadoop, Spark and other big data analytics platform for docking
l High compatibility, perfectly matching the mainstream CDN platform
S3 object storage using standard protocols, can be docked to an existing support smooth S3 protocol service system
Provide RESTFul API, console, and multilingual SDK toolkit
Provide development support services, support on-demand platform for rapid transformation of the line