Object-Based Redundancy for Metadata Recovery in Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data storage systems fail to effectively protect critical logical data structures, such as metadata and file systems, from secondary read failures, often resulting in catastrophic data loss, and are costly due to their reliance on sector-based redundancy schemes that do not differentiate between data types.
Innovation Solution
An object-based storage system that generates and stores redundancy data based on attributes of each data object, allowing for tailored protection levels and dynamic recovery of data objects using an object-based file system interface and controller, which can access and utilize redundancy data to reconstruct lost data without requiring complete hard disc module redundancy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional sector-based redundancy schemes are used, then all data is protected equally, but critical metadata and file systems remain vulnerable to catastrophic loss during secondary read failures
Solution Approach 1:
The patent applies different levels of redundancy protection to different data objects based on their importance. Critical data structures such as metadata and file system information receive stronger redundancy schemes, while less critical user data receives standard protection. This local differentiation ensures that catastrophic loss of critical information is prevented without uniformly over-protecting all data.
Solution Approach 2:
The patent segments data into distinct data objects with individual redundancy protection. Each data object is protected independently through object-based redundancy schemes, allowing selective application of protection levels. This segmentation enables the system to focus redundancy resources on critical metadata and file system structures rather than treating all sectors uniformly.
2Reliability
If RAID systems are used to protect data, then redundancy is provided across multiple discs, but the system incurs significant cost and performance penalties
Solution Approach 1:
The patent extracts the redundancy function from complex multi-disc RAID systems and implements it at the object level within a single storage device. By taking out the essential redundancy capability and applying it selectively to individual data objects, the system achieves data recovery protection without requiring multiple discs or incurring RAID-related cost and performance penalties.
Solution Approach 2:
The patent uses cost-effective object-based redundancy schemes that provide adequate protection without the expensive infrastructure of RAID systems. The redundancy data is stored efficiently alongside each data object, providing disposable-like replacement capability for failed data without requiring expensive redundant hardware.
3Reliability
If object-based redundancy protection is applied to all data objects, then critical data structures are protected, but storage space and write performance are reduced
Solution Approach 1:
The patent applies partial action by implementing object-based redundancy only for critical data structures such as metadata and file system information, rather than uniformly applying it to all data objects. This selective approach provides necessary protection for critical data while avoiding the performance penalty of generating redundancy for every single data object, thus maintaining better write performance.
Data Source
AI summary
A storage device comprising has a storage medium, a read-write mechanism, an object-based file system interface, and a controller. The read-write mechanism is adapted to read and to write data from and to the storage medium. The object-based file system interface within the storage device is adapted to organize and access data on the storage medium as objects and to access attributes of each data object. The controller is adapted to generate redundancy data for each data object to be stored on the storage medium according to the associated attributes and to store the data object and its associated redundancy data on the storage medium.