Limits and Notes
Limits
Product Type | Standard | | High-Performance | | Standard Turbo | | High-Performance Turbo | |
| Product Specifications | Recommended | Product Specifications | Recommended | Product Specifications | Recommended | Product Specifications | Recommended |
Maximum system capacity | 160TiB | 140TiB | 32TiB | 24TiB | 100PiB | 4PiB | 100PiB | 2PiB |
Minimum system capacity | None | | None | | 40TiB | | 20TiB | |
Maximum system bandwidth | 300MiB/s | 240MiB/s | 1GiB/s | 800MiB/s | 100GiB/s | 10GiB/S | 100GiB/s | 10GiB/s |
Maximum number of system files | Min[15,000 x used capacity (GiB), 1 billion] | Min[10,000 x used capacity (GiB), 0.8 billion] | Min[20,000 x used capacity (GiB), 1.5 billion] | Min[15,000 x used capacity (GiB), 1 billion] | Min[15,000 x deployed capacity (GiB), 1 billion] | Min[10,000 x deployed capacity (GiB), 0.8 billion] | Min[30,000 x deployed capacity (GiB), 1.5 billion] | Min[20,000 x deployed capacity (GiB), 1 billion] |
Maximum number of system directories | 10 million | 8 million | 15 million | 10 million | 10 million | 8 million | 15 million | 10 million |
Maximum length of filename | 255 bytes | 255 bytes | 255 bytes | 255 bytes | 255 bytes | 255 bytes | 255 bytes | 255 bytes |
Maximum length of absolute path | 4096 bytes | 4096 bytes | 4096 bytes | 4096 bytes | 4096 bytes | 4096 bytes | 4096 bytes | 4096 bytes |
Maximum number of directory levels | 1000 | 16 | 1000 | 16 | 1000 | 16 | 1000 | 16 |
Maximum number of files/subdirectories per directory | 1 million | 0.8 million | 1 million | 0.8 million | 1 million | 0.8 million | 1 million | 0.8 million |
Maximum number of concurrently opened files | 65536 | 1000 | 65536 | 1000 | 65536 | 1000 | 65536 | 1000 |
Maximum number of locks per file | 512 | 512 | 512 | 512 | 512 | 512 | 512 | 512 |
Maximum number of clients | 1000 | 100 | 1000 | 100 | 2000 | 1000 | 2000 | 1000 |
Maximum Bandwidth per Client | 300MiB/s | 300MiB/s | 500MiB/s | 500MiB/s | 10GiB/s | 10GiB/s | 10GiB/s | 10GiB/s |
Maximum number of mounted file systems per client | 1000 | 16 | 1000 | 16 | 16 | 8 | 16 | 8 |
Billing restrictions | Billed by the actual usage (excluding prepaid) | | Billed by the actual usage (excluding prepaid) | | Billed by the purchased capacity | | Billed by the purchased capacity | |
Supported protocol | NFS/SMB | | NFS | | POSIX/MPI | | | |
Supported OS | Linux/Windows | | | | Linux | | | |
Note
Turbo series
The Turbo series is mounted using a client. After you run the
mount command on the client installed, you can use the file system the same way as a local file system.The Turbo series operates and bills based on purchased capacity. For instance, if you purchase 40T of Standard Turbo, upon creation, regardless of the usage, you will be billed hourly for 40TiB. If used for an hour, the charge would be: 40*1024*0.6/24/30 = 34.13 yuan. If necessary, the file system can be destroyed at any time.
To ensure the cloud load balance of the file system after scaling up, we recommend that you scale up when around 80% of the capacity has been used. Online scale-up is supported and will be imperceptible during the whole process.
The Turbo series cannot be scaled down. You can create a Turbo instance, migrate your data, and then delete the old instance.
Because the self-deployed cluster needs to be set up again, the initial creation of the Turbo series will take about 20 minutes.
It is recommended to mount and use a Turbo file system only on a client in the same availability zone (AZ). Cross-AZ delays may cause issues such as client mount timeouts and disconnections.
Should you require a Turbo series with higher specifications (supporting a larger number of files, directories, etc.), please submit a ticket to get in touch with us.
UID and GID
When NFS v3.0 protocol is used, if UID or GID of the file does not exist in the local account, UID and GID will be displayed directly; otherwise, relevant user and group names will be displayed based on the mapping relationship of the local UID and GID.
When the NFS v4.0 protocol is used, if the Linux version is above 3.0, the UID rules and the GID rules will be the same as those of the NFS v3.0 protocol. Otherwise, the UID and the GID of all files will be displayed as
nobody.Note
When you mount a file system to a Linux version below 3.0 by using the NFS v4.0 protocol, we recommend that you refrain from performing change owner or change group on the file or directory. Otherwise, its UID and GID will become
nobody.Supported CIFS/SMB protocol
Supported protocol versions: CIFS/SMB 1.0 and later are supported. However, SMB 1.0 is not recommended for mounting, because it is inferior in terms of performance and features to SMB 2.0 and later and because Windows has stopped its technical support service for Windows versions supporting SMB 1.0 or earlier.
You cannot use NFS and SMB to access the same file system at the same time or directly access an SMB file system via WAN.
Read/write ACL is provided only at the file system level. No ACL is provided at the file/directory level.
IOCTL/FSCTL operations such as sparse files setting, file compression, ENI status query, and reparse point setting are not supported.
Alternate Data Streams are not supported.
Some protocol features in SMB 3.0 or later such as SMB Direct, SMB Multichannel, SMB Directory Leasing, and Persistent File Handle are not supported.