Peer Performance Model for Storage System Latency Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Latency measurements alone are not sufficient to provide a comprehensive assessment of storage system performance, as they can be influenced by external factors such as I/O request types and sizes, making them less informative for evaluating system performance effectively.
Innovation Solution
The implementation of a peer performance model and symptom model to assess storage system performance by normalizing latency variations and identifying underlying causes of performance issues through peer performance scores and symptom severity scores, respectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If latency measurements are used to assess storage system performance, then performance can be measured, but the measurements are influenced by external factors such as I/O request types and sizes making them less informative
Solution Approach 1:
The patent introduces peer storage systems as intermediaries to mediate the performance assessment. Instead of directly measuring raw latency which is contaminated by external factors, the system uses peer storage systems to provide comparable baseline measurements. This intermediary approach allows for normalized performance evaluation by comparing like-with-like operations across different systems, thereby recovering lost performance information while eliminating the distorting effect of external variables.
2Difficulty of detecting and measuring
If additional performance characteristics are monitored to improve diagnosis, then root cause identification improves, but system complexity increases
Solution Approach 1:
The patent segments the performance monitoring system into distinct functional components: latency measurement modules, peer comparison modules, symptom model modules, and diagnosis modules. Each component handles a specific aspect of performance assessment independently. This segmentation allows the system to monitor multiple performance characteristics without creating a monolithic complex system, as each segment can be developed, maintained, and scaled independently while contributing to the overall diagnostic capability.
Data Source
AI summary
An assessment of computing system performance may include evaluating, within a time window, the average latency of operations of a certain type and size with respect to a peer performance model constructed for operations of the same type and size. Such evaluation may result in the determination of a peer performance score. The peer performance score may be used to label performance characteristics of the computing system that are measured within the same time window. A library of such performance characteristics labeled with respective peer performance scores may be constructed by examining multiple computing systems, and such library may be used to construct one or more symptom models. Each symptom model may map a performance characteristic to a symptom severity score, which indicates the severity of a symptom of the computing system.