Operational Data Compression via Multi-Level Pattern Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current operational data compression methods in modern IT infrastructure yield low compression ratios due to reliance on general or record-based compression, leading to inefficient storage and energy utilization.
Innovation Solution
The method involves profiling operational data using normalizing functions, extracting patterns, and constructing dictionaries using machine learning, deep learning, and natural language processing techniques to compress data at line-level, sequence-level, and graph-level, thereby improving compression ratios and storage efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If general data compression or record-based compression methods are used, then the compression process is simple, but the compression ratio is low
Solution Approach 1:
The patent segments operational data into multiple hierarchical levels: line-level patterns (individual log entries), sequence-level patterns (temporal relationships between entries), and graph-level patterns (structural relationships between data elements). This multi-level segmentation enables progressive compression at each level, achieving high compression ratios while maintaining a systematic and manageable process
Solution Approach 2:
The patent introduces multiple dimensions of pattern recognition beyond traditional single-level compression. By analyzing data at line-level, sequence-level, and graph-level simultaneously, the system creates a multi-dimensional compression approach that captures relationships across different temporal and structural dimensions, significantly improving compression ratios
2Quantity of substance
If more storage space is allocated for operational data, then data retention is improved, but energy utilization deteriorates
Solution Approach 1:
The patent extracts and stores only the essential pattern information at each hierarchical level (line-level patterns, sequence-level patterns, graph-level patterns) rather than storing all raw operational data. This extraction approach maintains data retention capability while dramatically reducing storage requirements and associated energy consumption
Solution Approach 2:
The patent transforms operational data from its original high-volume format into compressed pattern representations with different parameters. By changing the representation parameters from raw data to extracted patterns, the system achieves efficient storage with reduced space and energy requirements while preserving essential information
3Speed
If traditional compression methods are used, then processing speed is maintained, but storage efficiency deteriorates
Solution Approach 1:
The patent performs preliminary pattern extraction and normalization operations on operational data before the actual compression process. By pre-processing data to identify and structure patterns at multiple levels beforehand, the system enables faster compression execution while achieving superior storage efficiency through the organized pattern representations
Data Source
AI summary
A method, computer system, and a computer program product for data compression is provided. The present invention may include receiving operational data. The present invention may include profiling the operational data using one or more normalizing functions. The present invention may include extracting a plurality of patterns from the operational data. The present invention may include compressing the operational data based on the plurality of patterns extracted.