Application of Anchao Distributed Storage in Non-Linear Editing System

With the wide application of digital technology in various industries, non-linear editing technology has also appeared around us and is developing rapidly. Non-linear editing is actually a way of arbitrarily arranging and editing video materials without following the original order and length. After the program is completed, you can change the length of one of the paragraphs or insert and delete other paragraphs. The application fields are mainly in professional video production fields such as animation production industry, radio and television news industry, small and medium multimedia studios.

In this issue, Zhihui Huayun specially invited Huayun Data Product Marketing Manager Zhang Kailiang to bring you the "Application of Anchao Distributed Storage in Non-Linear System".

Nowadays, with the improvement of the level of film and television production, higher requirements are continuously placed on back-end storage. For example, the number of workstations in the existing non-linear editing network continues to increase, and the resolution, code stream, and number of video materials continue to increase, making the performance and capacity of the original storage system unable to meet the demand. The following two issues need to be considered in the configuration storage process:

Performance requirements

The performance of the storage system should meet the current requirements for smooth operation of multiple non-linear editing workstations in the network. For the storage system, the total aggregate I/O bandwidth requirement is at least 10GB/s, which is mainly manifested as the read performance requirement. At the same time, video production is constrained by the frame rate and has a minimum requirement for delay. Therefore, the access delay of the storage system should be able to meet the requirements of the production software to work smoothly in the case of large concurrency, without obvious frame drop requirements.

Cross-platform data sharing

There are both Windows and Linux editing workstations in the user editing network, and data sharing between workstations is required. At the same time, it should also be able to interface with other office and business systems, such as media assets, film review, and daily office data storage. Therefore, storage systems must not only support a variety of mainstream operating systems, but also support common network file access protocols, such as: NFS, CIFS, FTP, etc. Various traditional applications based on the above-mentioned interfaces can be seamlessly migrated to the storage system to realize data sharing across interfaces and nodes.

system structure

Through the analysis and research of the above requirements, we have designed a non-woven network overall solution based on Anchao distributed storage system. The system is composed of multiple storage nodes and integrates two storage nodes of the existing storage system. The following figure shows the overall solution architecture of the upgraded storage system.

e879a0c0ac6d61bb6eb0b2216ac48ec2.jpeg

It can be seen from the system structure diagram that the Anchao distributed industrial storage system consists of four new storage nodes and two common components of the old equipment, which are connected to multiple Windows workstations, Mac workstations and office networks through 10 Gigabit switches. All workstations and office networks can access the storage system.

In this solution, nodes that require access performance such as editing workstations are all installed with the client (Windows/MacOS) of the Anchao distributed storage system. The PC in the office network can choose to use the general NAS protocol or client to access the Anchao distributed storage system according to its own situation.

Product Highlights

Rich access interface Anchao distributed storage supports users to access file data in NFS/CIFS/FTP mode. In addition, Anchao distributed storage also supports REST style Web Service interface (compatible with Amazon S3), HDFS interface (HDFS can be replaced in Hadoop environment)

Optimizing performance The performance advantages of Anchao distributed storage are manifested in two aspects. First of all, for a single client, Anchao distributed storage makes full use of the characteristics of data distributed storage, improves the access speed of a single client through concurrent access, and uses I/O optimization technologies such as caching, prefetching, and delayed writing to provide outstanding performance. Single client IO performance. Furthermore, for the aggregate performance of the entire system, Anchao distributed storage has good scalability, and the aggregate bandwidth and IOPS of the system can grow linearly with the growth of the number of nodes. System aggregation performance that is difficult to obtain with traditional storage can be obtained. For example, the aggregate I/O bandwidth can reach tens of GB/s, and the aggregate IOPS can reach millions.


Guess you like

Origin blog.51cto.com/14987653/2546193