Copy-less Database Transaction File Restore via Hardlinks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for backing up and restoring transaction files in database storage systems require significant resources and time due to the need for copying files, leading to intensive input/output operations and substantial storage space consumption.
Innovation Solution
A mechanism that archives and restores transaction files without copying them by creating hardlinks in an archive directory, allowing for a 'copy-less' backup and restore process, which reduces disk space consumption and server workload by using existing file system pointers to manage file references.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional backup methods copying transaction files are used, then complete backup of transaction files is achieved, but disk space consumption and server workload increase significantly
Solution Approach 1:
The patent applies the copying principle by creating hard links (a form of file system pointer) that reference the original transaction files without duplicating the actual file data. This allows the backup system to maintain complete transaction file backups while consuming minimal additional disk space, as hard links use metadata pointers rather than duplicating the entire file content.
Solution Approach 2:
The patent introduces hard links as an intermediary mechanism between the original transaction files and the backup storage. These hard links act as file system pointers that reference the original files, enabling the backup system to access and preserve transaction file integrity without directly copying the file data, thus reducing disk space consumption and server workload.
2Reliability
If conventional backup methods copying transaction files are used, then complete backup of transaction files is achieved, but backup and restore time increase due to intensive I/O operations
Solution Approach 1:
The patent uses hard links (a pointer-based copying mechanism) to create backup references to transaction files without performing actual file data copying. This dramatically reduces I/O operations during backup, as the system only needs to create metadata pointers rather than transfer entire file contents, thereby reducing backup time while maintaining backup completeness.
Solution Approach 2:
The patent performs preliminary creation of hard links during the backup process, establishing file references before actual backup operations are needed. This preliminary action of creating pointers rather than copying data reduces the time required for both backup and subsequent restore operations, as the file system can quickly resolve these pointers without performing extensive data I/O.
3Reliability
If transaction files are copied to archive directory, then backup of transaction files is achieved, but server workload and storage subsystem I/O operations increase
Solution Approach 1:
The patent implements a pointer-based copying mechanism using hard links instead of actual file copying. This allows the backup system to achieve complete transaction file backups while minimizing server workload, as creating hard links is a lightweight metadata operation that does not require intensive data processing or transfer, unlike conventional file copying methods.
Data Source
AI summary
A mechanism is provided for archiving and restoring transaction files of a database system without requiring the copying of the transaction files. A server system executes a host application that implements the database system comprising a set of database files and a set of transaction files that contains updates to the set of database files, the files being stored in a storage system. A backup engine interacts with the host application to archive a transaction file having an original filename in an original directory by creating a corresponding archive filename to the transaction file in an archive directory. The archive transaction files may be used later to perform a copy-less restore process of the database system that does not require copying of any transaction files. The copy-less restore process restores any altered original filenames by replacing, in the original directory, an altered original filename with its corresponding archive filename.