Shadowed File System Lock-Coupled Traversal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed file systems face delays due to the need for write access to persistent memory when changes are made to file objects, causing concurrent access issues and delays in read and write requests.
Innovation Solution
A shadowed file system with lock-coupled namespace traversal that maintains object attributes by using a tree structure and propagates rolling updates through a notification list, allowing updates to be made in parallel with object data and attributes, and employing object versioning to manage changes across transactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If changes to file objects are written to persistent memory in commit order, then data integrity is maintained, but significant delays occur in subsequent read and write requests
Solution Approach 1:
The patent applies preliminary action by maintaining a shadowed file system that pre-organizes object attributes by category (physical size, logical size, etc.) before actual commits occur. This allows the system to prepare attribute update structures in advance, so when commits happen, the attribute updates can be propagated immediately without waiting for persistent memory writes to complete.
Solution Approach 2:
The patent introduces an intermediary mechanism through the shadowed file system that acts as a buffer between object changes and persistent memory commits. The shadowed file system maintains attribute information in memory organized by category, allowing read requests to access attribute data without waiting for persistent memory writes, while still ensuring integrity through the commit process.
2Reliability
If locking mechanisms are used to maintain data consistency during concurrent access, then data integrity is ensured, but system performance and responsiveness deteriorate
Solution Approach 1:
The patent applies segmentation by dividing object attributes into distinct categories (physical size, logical size, and other attributes). Each category can be updated and accessed independently through the shadowed file system. This segmentation allows concurrent access to different attribute categories without requiring global locks, improving system performance while maintaining consistency within each category.
Solution Approach 2:
The patent implements dynamics through the shadowed file system that dynamically manages attribute updates without static locking. The system uses a notification list mechanism that allows multiple transactions to access and update attributes concurrently, with changes propagated dynamically as they occur rather than being blocked by locks. This dynamic approach maintains data consistency while improving responsiveness.
3Reliability
If attribute updates are performed sequentially to ensure consistency, then data integrity is maintained, but concurrent access capability is reduced
Solution Approach 1:
The patent applies universality through the shadowed file system that serves multiple functions simultaneously: it maintains attribute consistency, enables concurrent access, and propagates updates across multiple transactions. The notification list mechanism allows the same attribute update structure to serve multiple transactions concurrently, eliminating the need for sequential processing while maintaining consistency through the universal commit process.
Data Source
AI summary
A deferred mechanism of rolling attribute changes to objects at transaction commit time is described. To support rolling updates, attributes are categorized into three types—increment, decrement or replace. Since transactions are committed in order, the update process uses a notification list of dependent transactions to propagate attribute changes as a function of attribute category or type. Shadow based file-systems (with or without object versioning) realize efficiency gains by employing the deferred mechanism.