xfs_estimate

xfs_estimate [ options ] mount_point ...

-b average_bytes
    Specifies the average size of files in bytes to use for the estimation. If omitted, the minimum file size (typically one block) is assumed.

-i inodes
    Specifies the total number of inodes (and implicitly files) to use for the estimation. This option implies -f.

-f files
    Specifies the total number of files to use for the estimation. If neither -i nor -f is given, the current number of files on the filesystem is used.

-m
    Estimates space required for a mirrored filesystem setup, assuming each data block consumes twice the space.

-L
    Optimizes the estimation for very large directories, where directory block allocation patterns might differ.

-v
    Enables verbose output, displaying more details about the estimation process and results.

-q
    Suppresses most output, showing only essential information (quiet mode).

-d
    Enables debug output, which implies verbose mode and provides even more detailed internal information.

mount_point ...
    One or more mount points of the XFS filesystems to estimate space for.

xfs_estimate is a Linux command-line utility designed to predict the amount of disk space an XFS filesystem will consume under specific conditions.
It is particularly useful for capacity planning, allowing administrators to estimate storage requirements before populating a filesystem with data. The tool takes into account various XFS-specific overheads, such as inode allocation, directory structures, and block sizes, which are crucial for workloads involving numerous small files where these overheads can be significant.
Users can specify the desired number of files or inodes and an average file size. Based on these inputs and the underlying XFS filesystem's characteristics (like block size and inode size), xfs_estimate calculates an approximate total space requirement, including both data and metadata overhead.
This command provides a valuable insight into how different file counts and sizes will impact disk consumption, helping to optimize filesystem sizing and avoid unexpected out-of-space scenarios.

The output of xfs_estimate is an approximation and should not be treated as an exact calculation. It assumes a uniform distribution of file sizes (when -b is used) and does not account for potential fragmentation or specific application-level block allocation patterns that might deviate from general XFS behavior. It is designed for XFS filesystems only.

xfs_estimate calculates space by considering data blocks (based on average file size and count), inode blocks (based on inode count and inode size), and directory blocks (based on file count and directory block size). It uses filesystem metadata obtained from the mounted XFS filesystem to determine block sizes, inode sizes, and other structural parameters.

This command is invaluable for administrators setting up new XFS filesystems for applications that generate a large number of small files (e.g., mail servers, web caches, version control systems), where metadata overhead can be a primary concern. It helps in determining the optimal size for a new filesystem or whether an existing filesystem can accommodate a projected increase in data.

xfs_estimate is part of the XFS user-space utilities, which were originally developed by Silicon Graphics, Inc. (SGI) for their IRIX operating system. XFS, known for its high performance and scalability, was later ported to Linux. This utility was created to assist administrators in capacity planning, addressing the specific challenges of XFS's advanced allocation mechanisms, particularly the overhead associated with managing millions of inodes and directory blocks, which can consume significant space on large filesystems.

mkfs.xfs(8), xfs_info(8), xfs_growfs(8)