The Cloud File Service provides types of standard and capacity. The standard one provides stable highly shared bandwidth throughput and IOPS performance, which is applicable to frequent read/write scenarios; the capacity one is applicable to non-frequent read/write scenarios, such as filing and backup, which provides shared file storage of better cost performance.
The Cloud File Service supports NFS v4.0 and NFS v4.1 protocols and can attach by using standard file system attach command.
Multiple Virtual Machines can simultaneously access file storage created in the Cloud File Service, conduct read/write operation of files and implement data share of multiple computing nodes through NFS protocol.
It is unnecessary to preset the file storage capacity after you creating the file storage. The file storage capacity can implement auto scaling according to your operation of adding or deleting files therein. You only need to pay for the actual usage of file storage.
The Cloud File Service adopts distributed architecture with three-copy design. All files and directories of user are dispersed over different fault domain storage to prevent the single-point fault from causing data inaccessibility or data loss. It has high availability and high persistence.
The mount points of the Cloud File Service are set in user's VPC, so all accesses to file storage are protected by the network security isolation control of user's VPC; and it supports users to control read/write access to file data through standard POSIX permission control.
The rendering users can upload the rendering materials to the Cloud File Service to form a rendering task material library. Rendering clusters consisting of CPUs or GPU Virtual Machines read the required material files respectively from the Cloud File Service to conduct rendering. Cache the intermediate result of the rendering task in the Cloud Disk Service of the local machine or in the Local Disk. After the rendering task is completed, output the rendering result to the Object Storage Service for subsequent business use.
The AI platform stores various training data and models in one or more file storages respectively. AI training clusters consisting of Virtual Machines read model sand training sets required for the training task from the Cloud File Service respectively for training. Output the final training result to the Object Storage Service for subsequent business use.
In media business, files, images, graphs, audios and videos shall be provided with shared storage and edit processing. The high throughput and shared file access functions of the Cloud File Service can provide a unified bucket for media business for all users to improve the processing efficiency of media business.