PowerScale OneFS 9.5.0.0 Technical Specifications Guide

PDF

File system guidelines

This section presents guidelines for configuring the OneFS file system.

For assistance, contact your PowerScale account representative or PowerScale Technical Support.

Table 1. OneFS file system specifications
Item	OneFS 9.5.0.0	Description
Block size	8 KB	The maximum block size limit. This limit cannot be changed.
Cluster name length	40 characters	The maximum length for the cluster name.
Cluster size	252 nodes	The maximum number of nodes that a cluster can have.
Custom access pattern templates	5	The limit for custom file-system-tunable templates. This limit is in addition to the default templates of "random," "streaming," and "default."
Directories per directory	100,000	The recommended limit for the number of directories in a directory. Exceeding this limit might negatively affect the cluster performance and client connections. Evaluate the workflow and workloads for your cluster to determine the value that works best for your environment.
Directory depth	509	Maximum recommended depth of a directory tree is 509.
FEC protection	+4n	The following FEC options are supported: +1n, +2n, +2d:1n, +3d:1n, +3d:1n1d, +3n, +4d:1n, +4d:2n, +4n. OneFS protection is defined at the node pool level. A cluster with multiple node pools can have multiple protection schemes simultaneously. The recommended protection level depends on the size of the node pool and node types. For information about disk pools, node pools, and tiers, see the white paper Storage Tiering with Dell EMC Isilon SmartPools.
Mirrored protection	8x (maximum)	Mirroring options from 2x to 8x are supported. The recommended value depends on the node pool size.
File clones per file	32,766	The maximum number of references for a single block in a shadow store. When the limit for file clones per file is exceeded, a new shadow store is created.
File name length	Up to 1024 unicode characters in namelength domains. 1024 bytes in regular directories.	In namelength domains, OneFS can support upto 1024 unicode characters. In regular directories, OneFS supports a maximum filename length of 1024 bytes. Most Unicode character encodings, such as UTF-8, specify that a character can have multiple bytes. UTF-8 can have up to 4 bytes per characters. The characters in some languages, such as Japanese, are likely to have multiple bytes per character. OneFS supports UTF-8 by default.
Standard file size	4.398 TB (4 TiB)	The maximum size for a file for all PowerScale clusters. Files larger than 1 TB can negatively affect job engine performance.
Expanded file size	17.6 TB (16 TiB)	The maximum size for a file that can be supported with specific PowerScale hardware configurations.
File system size	181.44PB	The maximum capacity for the file system. The capacity size does not include overhead for the OneFS operating system, the file system, or data protection.
Files per directory	1,000,000	The recommended limit for files per directory. Exceeding this limit might negatively affect the cluster performance and client connections. Evaluate the workflow and workloads for your cluster to determine the value that works best for your environment. To improve performance when managing large numbers of files in a single directory, use nodes that have solid-state drives (SSDs).
Hard links per file	1,000	The default and maximum number of hard links per file. You can set the maximum number of hard links per file with the efs.ifm.max_links system control. Setting the number higher than the default value can slow snapshot operations and file deletions. For more information, see the EMC Isilon knowledge base article 447064, OneFS: Sysctl: efs.ifm.max_links.
Inodes per cluster	Billions	OneFS dynamically allocates new inodes from free file system blocks. The maximum number of possible inodes depends on the number and density of nodes in the cluster, as expressed by the following formulas: 4Kn drives: ((number of nodes in the cluster) * (node raw TB) * 1000^4 * .99) / (8192 * (number of inode mirrors)) 512n drives: ((number of nodes in the cluster) * (node raw TB) * 1000^4 * .73) / (512 * (number of inode mirrors)) See the guideline for files per directory. The limit for files per directory can limit the number of files that the system can store.
Logical Node Numbers (LNNs)	252	The limit for logical node numbers.
Node pools per cluster	20	The recommended and maximum limits for node pools per cluster. The number of node pools that can be created is limited by the number of nodes in the cluster.
Open files per node	315,000	The maximum number of open files per node depends on the maximum number of vnodes that are available on the node. The amount of available vnodes depends on how much RAM the node has. The maximum number of open files per node is 90% of the maximum number of vnodes on that node, as expressed in the following formula: kern.maxfiles = kern.maxvnodes * .9 The OneFS protocol daemons, such as the input-output daemon (lwio), might impose additional constraints on the number of files that a node can have open. The protocol daemons typically impose such constraints because the kernel places limits on per-process memory consumption.
Path length	4096 bytes	The maximum length for a pathname. The length is the maximum length of a directory path that can be passed into a system call; it does not represent the absolute depth of nested directories. Shorter path and file names require fewer lookup operations. As a best practice, keep your path and file names as short as possible, especially in workflows that include many lookups. OneFS features like NDMP and SyncIQ may not work as expected on paths longer than the maximum limit. NOTE: For symbolic links, the path length of the target is restricted to 1024 bytes if the symlink source is in a restricted domain.
Device IDs	65,535	Device IDs are unique identifiers for nodes. Device IDs are not reused when nodes are replaced. To reach the limit of Device IDs in a three-node cluster, you must replace nodes 65,532 times.
User attribute keys	16	The limit of attribute keys that can be created within any file system object. The user attribute term refers to custom file system metadata that the FreeBSD extattr API creates. These extended attribute data types can be acted on by SmartPools, for example, by choosing the File Attribute file pool policy filter. Extended attributes exist as "name=value" pairs within a file system object.
User attribute key size	24 bytes	The limit size for the user attribute key.
User attribute value size	128 bytes	The limit size for the user attribute value.
User attribute total size	1 KB	The limit for the size of the custom metadata that is associated with the file system object.