Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.Original topic: 文件系统要求
|
username: Clover
Is there anything unsuitable about XFS? Can anyone explain?
|
username: Clover
|
username: It is said that XFS is not as stable as EXT4 and has performance fluctuations.
|
username: In fact, both can be used. The official recommendation is as follows: based on our production environment practice, it is best to use SSD physical disks, as cloud disk IO is not sufficient.
|
username: 
|
username: xfs: Supports single volumes up to 8EiB, with a maximum file size of 8EiB. It more efficiently supports parallel IO operations, has better scalability on RAID, and handles high multi-threaded concurrent pressure better than ext4. However, it cannot compress and performs poorly when deleting a large number of files. It is better suited for scenarios with high demands for large file read/write and throughput, and has strong scalability for handling large files.
ext4: Maximum volume of 1EiB and maximum file size of 16TiB. It has mechanisms like journal checksums and delayed allocation, which can better protect data integrity and consistency. It is better suited for scenarios with high demands for file deletion and fault tolerance, and is more efficient at handling small files.
|
username: xfs is also an option, but the official recommendation is to use ext4.
Some previous posts mentioned that XFS has a risk of data loss. To be safe, it’s better to use ext4 in a production environment.
|
username: 
|
username: 
|
username: In fact, both can be used. If you are pursuing extreme performance, use ext4. In high concurrent read and write scenarios, the xfs file system can have a certain impact on the performance of NVMe SSDs.
|
username: This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.