Hierarchical Storage File Reclassification via Stub Metadata
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Hierarchical Storage Management (HSM) systems face challenges in reclassifying files efficiently, leading to slow responses during file recall due to ongoing file access, as they require exclusive recall of files for reclassification, which can occupy tape drives and cause delays in accessing files.
Innovation Solution
A method that determines the necessity of reclassification based on stub metadata, updates the classification using the latest model, and migrates files from primary to secondary storage, allowing reclassification without preventing file access, by storing model information in file stubs and only recalling files with outdated models for reclassification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If files are recalled for reclassification in existing HSM systems, then reclassification can be performed, but file access is blocked and tape drives are occupied causing delays
Solution Approach 1:
The system performs preliminary checks by examining stub metadata to determine whether reclassification is necessary before initiating file recall. This preliminary action filters out files that do not require reclassification, preventing unnecessary tape drive occupation and file access delays while ensuring that only files needing reclassification are recalled.
Solution Approach 2:
The system uses stub metadata that already contains classification model information to self-determine whether reclassification is needed. This self-service mechanism eliminates the need to recall all files for reclassification, allowing continuous file access while maintaining reclassification accuracy for files that actually need it.
2Reliability
If all files are recalled for reclassification, then reclassification completeness is improved, but tape drive availability decreases and access speed worsens
Solution Approach 1:
Instead of recalling all files for reclassification, the system performs partial action by only recalling files that meet specific criteria (outdated classification models) as determined by stub metadata analysis. This partial action maintains reclassification completeness for files that need it while preserving fast access speeds for files that don't require reclassification.
Solution Approach 2:
The system applies different treatment to different files based on their individual characteristics stored in stub metadata. Files with outdated classification models are recalled for reclassification, while files with current models are left untouched. This local quality approach ensures reclassification completeness where needed while maintaining high access speeds overall.
3Measurement precision
If file recall is performed for reclassification, then classification accuracy can be updated, but system productivity decreases due to tape drive occupation
Solution Approach 1:
The system performs preliminary analysis of stub metadata to identify files requiring reclassification based on outdated classification models. This preliminary action enables targeted recall of only necessary files, updating classification accuracy for those files while maintaining high system throughput by avoiding unnecessary tape drive occupation.
Solution Approach 2:
The system uses feedback from stub metadata containing classification model information to determine whether reclassification is needed. This feedback mechanism ensures classification accuracy is updated for files with outdated models while preserving system productivity by filtering out files that already have current classifications.
Data Source
AI summary
In an approach to automatically reclassifying files in an HSM system, responsive to occurrence of a recall to move a file on a hierarchical storage system from a secondary storage to a primary storage, whether stub metadata exists for the file is determined. Responsive to determining that stub metadata exists for the file, whether reclassification of the file is necessary is determined based on the stub metadata. Responsive to determining that the reclassification of the file is necessary, the file is reclassified, where the reclassification uses a latest classification model, and further where the stub metadata is updated with results of the reclassification. The file is migrated from the primary storage to the secondary storage based on the stub metadata.