Sparse Backup Extent Mapping for File System I/O Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing backup methods for file systems either significantly impact file system performance or unnecessarily reduce network performance by copying all data, including unused portions, during the backup process.
Innovation Solution
Implementing a sparse backup system that identifies and copies only unallocated extents from a primary storage device to a backup storage device, using a mapping to translate accesses and store data efficiently, thereby reducing unnecessary data transfer and maintaining file system performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If backups are performed by copying individual files through the file system, then complete file backup is achieved, but file system performance deteriorates due to increased I/O load and computation
Solution Approach 1:
The storage volume is divided into allocated and unallocated extents. The backup process segments the copying operation to only include allocated extents that contain actual file data, excluding unallocated extents. This segmentation allows complete backup of used data while avoiding unnecessary processing of unused space, thus maintaining file system performance.
Solution Approach 2:
The invention extracts and identifies allocated extents from the total storage volume using file system metadata (such as inode tables or master file tables). By taking out only the allocated portions that contain actual file data for backup, the system avoids the performance degradation caused by processing the entire volume including unallocated space.
2Reliability
If the entire storage volume is copied for backup, then all data is backed up, but network performance decreases due to copying needless unallocated data
Solution Approach 1:
Instead of performing a complete volume copy, the invention applies partial action by copying only the allocated extents that contain actual file data. The file system metadata is queried to identify which extents are allocated, and backup operations are performed only on those specific portions, reducing network bandwidth consumption while ensuring all actual data is backed up.
3Measurement precision
If metadata is accessed to locate files for backup, then accurate file location is achieved, but I/O operations increase due to accessing primary storage media
Solution Approach 1:
The file system metadata (inode tables, master file tables) is accessed preliminarily during the backup initiation phase to identify and map all allocated extents before the actual data copying begins. This preliminary action caches the extent location information in memory, allowing the subsequent backup process to proceed without repeated I/O operations to locate files, thus reducing overall I/O time while maintaining accurate file location.
Data Source
AI summary
Systems and methods for performing sparse backups are disclosed. One method may involve: receiving information indicating unallocated extents from a file system that organizes data on a primary storage device; copying to a backup storage device data from extents of the primary storage device, where the unallocated extents indicated by the information are not copied; and for each of a plurality of copied extents, storing a mapping from an original extent on the primary storage device to a new extent on the backup storage device into which data stored in that original extent is copied. A backup volume that includes the data included in plurality of copied extents may be mounted, and the mapping may be used to translate accesses to the extents on the backup volume to extents on the backup storage device.