Overload Correlation for Storage Throughput Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Previous data storage systems fail to effectively address overloading of shared hardware components, leading to missed response time service level objectives (SLOs) due to the lack of performance data from individual hardware components, which prevents adequate performance control application.
Innovation Solution
The system monitors performance indicators for units of managed storage objects (UMOs) and generates overload correlations between UMOs that share hardware components, applying performance controls to competing UMOs to bring the target UMO's performance within acceptable ranges by adjusting throughput and response times based on calculated overload correlations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If performance controls are applied based on individual hardware component monitoring, then component overloading can be detected and addressed, but performance data from individual hardware components is unavailable preventing adequate performance control application
Solution Approach 1:
The patent combines performance data from multiple storage objects into aggregate performance data for shared hardware components. By merging the performance metrics of multiple storage objects that share the same hardware components, the system reconstructs the performance information that would otherwise be unavailable from individual component monitoring, enabling effective performance control application.
2Productivity
If shared hardware components are utilized by multiple storage objects, then resource utilization efficiency is improved, but overloading of shared hardware components causes storage objects to miss response time SLOs
Solution Approach 1:
The patent implements a feedback mechanism that continuously monitors aggregate performance data of shared hardware components and adjusts performance controls dynamically. When overloading is detected through aggregate performance analysis, the system applies performance controls to affected storage objects to bring them back into SLO compliance, while maintaining efficient resource utilization by only intervening when necessary.
Solution Approach 2:
The patent changes performance parameters of storage objects dynamically based on detected overloading conditions. By adjusting parameters such as I/O rates, prioritization levels, or throttling settings of storage objects sharing overloaded hardware components, the system maintains both high resource utilization and SLO compliance through adaptive parameter modification.
3Reliability
If performance controls are applied to balance workload across hardware components, then component overloading frequency is minimized, but performance controls cannot be effectively applied when performance data is unavailable
Solution Approach 1:
The patent merges performance data from multiple storage objects to reconstruct component-level performance information. By combining aggregate performance metrics from multiple storage objects that share hardware components, the system creates sufficient statistics to detect overloading conditions and apply workload balancing controls without requiring direct access to individual component performance data.
Data Source
AI summary
Improved techniques for applying performance controls in a data storage system based on overload correlations between units of managed storage objects (UMOs). When a performance indicator (e.g. response time) for a target UMO is outside an acceptable range, a competing UMO that potentially shares at least one potentially shared hardware component with the target UMO is identified. An overload correlation between the target UMO and the competing UMO is generated that indicates an amount of correlation between the performance indicator for the target UMO and a performance indicator for the competing UMO. A performance control is then applied to the throughput of the competing UMO that reduces the throughput of the competing UMO. The amount that the throughput of the competing UMO is reduced is based on the amount of overload correlation between the target UMO and the competing UMO.