Storage System Trace-Based Management for Parameter Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage systems lack sufficient statistical measurements to effectively modify parameter settings, leading to inadequate decision-making for performance optimization.
Innovation Solution
A method and system that record I/O transactions into an event log, process them using a summarizer to generate statistical measurements, and employ these measurements to support parameter setting decisions, allowing for adjustments without disrupting system operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If aggregated statistical measurements are used to determine system characteristics, then parameter setting decisions can be made, but the measurements are insufficient for making appropriate decisions for parameter modifications
Solution Approach 1:
The patent segments the measurement system into two distinct components: an event collection module that gathers detailed I/O operation data, and an event processing module that transforms this data into meaningful statistics. This segmentation allows the system to maintain both high measurement precision through detailed event tracking and manageable complexity through modular processing architecture.
2Loss of information
If detailed I/O transaction data is collected and processed, then accurate statistical measurements can be generated, but the processing complexity and resource requirements increase
Solution Approach 1:
The patent extracts only the essential and relevant features from detailed I/O transaction data to generate statistical measurements. The event processing module selectively extracts meaningful metrics (such as I/O throughput, latency, and operation patterns) while discarding redundant information, thereby maintaining information completeness for decision-making while reducing processing complexity.
3Adaptability or versatility
If parameter settings are modified to accommodate customer requirements, then system adaptability improves, but system stability and performance consistency may be affected
Solution Approach 1:
The patent implements a feedback mechanism where statistical measurements derived from detailed event data provide continuous information about system performance and workload characteristics. This feedback enables informed parameter adjustments that adapt to customer requirements while maintaining system stability, as decisions are based on actual measured performance rather than arbitrary changes.
Data Source
AI summary
Embodiments of the invention relate to storage systems, and for recording event records into a log. Different statistics may be tallied from the log, with the different statistics generating different statistical measurements. The aspect of processing of I/O and preparation of statistical measurements are separated, thereby enabling data structures and algorithms which would not be considered feasible for use as part of I/O processing, because they may compromise the performance or other essential characteristic of I/O processing, to be employed.