Storage Dependency Mining via Search Tree Pruning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data prefetching and caching techniques in storage systems are inefficient and complex, requiring improved dependency mining methods to optimize data management based on access patterns.
Innovation Solution
A method involving determining block data source sequences, constructing search trees based on activity patterns, and performing search tasks using block dependency criteria to identify dependent blocks, with the option to aggregate results from multiple search trees and prune search tree branches, is implemented to enhance storage management processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex data mining algorithms are used for dependency mining, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent segments the data mining process into distinct phases: determining block data source sequences, splitting sequences into subsequences, determining activity patterns, constructing search trees, and performing search tasks. This segmentation transforms a complex monolithic algorithm into manageable modular components, reducing overall system complexity while maintaining mining accuracy
Solution Approach 2:
The patent introduces search trees as an intermediary data structure between raw access sequences and dependency mining results. The search trees organize activity patterns in a hierarchical structure that facilitates efficient searching and pruning, acting as a mediator that simplifies the computational process while preserving measurement precision
2Reliability
If comprehensive dependency mining is performed, then reliability is improved, but loss of time increases
Solution Approach 1:
The patent performs preliminary actions by determining activity patterns and constructing search trees before executing the actual dependency mining search task. This preliminary organization of data into structured search trees enables faster retrieval and analysis during the mining process, reducing overall execution time while maintaining comprehensive analysis
Solution Approach 2:
The patent implements pruning mechanisms that perform partial action by selectively exploring only the most promising branches of the search tree based on block dependency criteria. This avoids exhaustive search of all possible paths, significantly reducing mining time while still identifying critical dependencies with high reliability
3Measurement precision
If multiple search trees are constructed, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent merges multiple search trees into a unified search framework that processes different subsequences and activity patterns simultaneously. By combining multiple search trees and aggregating their results, the system achieves more precise dependency mining while managing complexity through a standardized search interface and result aggregation mechanism
Data Source
AI summary
A system and method is provided in which access patterns of data blocks are used to predict future accesses to the data blocks. One of the types of patterns that may be used in connection with the system described herein is depended blocks. Dependent blocks are blocks that are often referenced one after another in a short period of time. For example, one block (block b) is commonly referenced after another block (block a) a noted time period. The system described herein provides a block dependency algorithm that enables a search task involving searching of block dependencies with respect to proposed dependency definitions. In an embodiment, the search task may be performed based on using of a branch and bound methodology. A search tree is built in which a set of activity vectors (V) are ordered, and a search algorithm is used to traverse and “prune” branches of the tree.