File System Hot Cold Classification via Write Pattern Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for classifying hot and cold files based on predefined rules face limitations, particularly when dealing with files having unique extensions or no extensions, and ambiguous directory names.
Innovation Solution
An electronic device that includes a processor configured to write application files in memory, identify and update their write patterns, classify files as hot or cold based on these patterns, and store classification results with the files, utilizing machine learning to assign weights to features of the write patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If predefined rules (file extension, directory name) are used to classify hot/cold files, then the classification process is simple, but the prediction accuracy deteriorates when files have unique extensions or ambiguous directory names
Solution Approach 1:
The patent transitions from using static parameters (file extension, directory name) to dynamic parameters (write pattern features such as write frequency, write size, and write interval) for file classification. This allows the system to adapt to varying file characteristics and improve prediction accuracy without significantly increasing system complexity.
Solution Approach 2:
The patent replaces the mechanical rule-based classification system with a machine learning-based system that automatically learns optimal classification criteria from actual write patterns. This substitution enables accurate prediction of file types even when traditional rules fail, while the learned model can be efficiently implemented in software.
2Measurement precision
If machine learning is used to learn write patterns and classify files, then the file type prediction accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent implements a hybrid approach where machine learning is used selectively - the processor learns write patterns for files and uses this learning for classification, but the overall system maintains the existing file system structure and storage mechanisms. This partial application of machine learning provides accuracy improvements without requiring complete system redesign.
Solution Approach 2:
The system performs self-learning of write patterns automatically in the background without requiring external intervention or complex configuration. The processor continuously monitors and learns from actual file write operations, adapting the classification model autonomously while maintaining normal system operations.
3Productivity
If hot and cold files are separated and stored in different areas, then storage performance is optimized and storage lifespan is extended, but the device complexity increases
Solution Approach 1:
The patent divides the storage into different areas for hot files and cold files based on their access patterns. This segmentation allows optimized storage operations - frequent writes for hot files and bulk operations for cold files - improving overall storage performance and extending NAND flash lifespan through reduced write amplification.
Solution Approach 2:
The system dynamically classifies files as hot or cold based on learned write patterns rather than using static classification rules. This dynamic approach allows the storage management strategy to adapt to changing file characteristics and access patterns, optimizing performance automatically without manual intervention or complex configuration.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
According to various embodiments, an electronic device may include a random access memory, a storage, and a processor. The processor may be configured to write a file of an application in the memory in response to a file input request of the application, to monitor a write pattern of the file at a first time of writing the file of the application in the memory, followed by updating the write pattern in the memory, to classify the file as one of a hot file and a cold file based on the write pattern of the file at a second time of copying the file written in the memory to the storage, and to control storing a classification result of the file together with the file in the storage or storing the file in a first area or a second area of the storage based on the classification result of the file. The hot file is a file whose modification and/or deletion occurs frequently compared to the cold file, and the cold file is a file whose modification and/or deletion occurs infrequently compared to the hot file. The write pattern may include a degree of modification of the file.