Disk Detection Using Multi-Counter I/O Response Time Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing disk detection methods are inadequate in accurately identifying slow disks, leading to performance degradation and potential service interruptions in RAID and distributed storage systems due to mechanical or environmental issues, as they fail to comprehensively assess real-time I/O response times and related counter data.

Innovation Solution

A method and apparatus that collect real-time data from multiple I/O-related counters, including I/O response time and other affecting counters, to determine abnormal response times by referencing preset intervals, clustering centers, and load ranges, thereby providing a more accurate detection of slow disks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional single-counter detection methods are used, then the detection process is simple, but the detection accuracy is low leading to false negatives and false positives

Engineering Contradiction:
Improvedetection accuracyVSAvoiddetection method complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The detection method segments the I/O response time evaluation into multiple independent counter measurements (read response time, write response time, read I/O operations, write I/O operations). Each counter is evaluated separately against its own threshold, and the final detection result is determined by combining these segmented evaluations. This segmentation allows comprehensive assessment without relying on a single complex metric.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The invention transitions from single-dimensional detection (one counter) to multi-dimensional detection (multiple counters). By introducing additional detection dimensions (different I/O operations and response time measurements), the system achieves more accurate disk health assessment. The multi-dimensional approach captures the complex nature of disk performance that cannot be reflected by a single counter.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Reliability

If multiple I/O-related counters are monitored, then the detection accuracy improves, but the detection complexity increases

Engineering Contradiction:
Improvesystem reliabilityVSAvoiddetection system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The monitoring system segments the overall detection task into independent counter monitoring sub-tasks. Each I/O-related counter (read response time, write response time, read I/O operations, write I/O operations) is monitored independently with its own threshold. This segmentation maintains system reliability through comprehensive monitoring while managing complexity by keeping each sub-task simple and independent.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system implements feedback mechanisms where each counter's current value is continuously compared against its threshold, and the detection result is adjusted based on the feedback from multiple counters. The final detection decision incorporates feedback from all monitored counters, ensuring reliable detection while systematically managing the complexity through structured feedback integration.

Inventive Principle:
Principle #23Feedback

3Measurement precision

If comprehensive I/O counter data is collected, then false negatives and false positives are reduced, but the data processing load increases

Engineering Contradiction:
Improvedetection precisionVSAvoiddata processing efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The data collection and processing is segmented into independent counter evaluations. Each I/O counter is processed separately against its threshold, producing individual evaluation results. This segmentation reduces the computational complexity of data processing while maintaining high detection precision through comprehensive multi-counter analysis.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system monitors more counters than the minimum single counter, implementing a form of excessive action that collects more data than strictly necessary. This partial/excessive monitoring approach improves detection precision by capturing multiple aspects of disk performance, while the independent threshold comparison method keeps the processing load manageable.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3321807B1Disk detection method and device
Publication Date: 2019.09.18 HUAWEI TECH CO LTD
  • EP3321807B1 patent drawingFigure 1
  • EP3321807B1 patent drawingFigure 2
  • EP3321807B1 patent drawingFigure 3~4

AI summary

The present invention provides a disk detection method and apparatus. The method includes: collecting a set of N pieces of real-time data that are in a one-to-one correspondence with N input/output I/O-related counters of a disk, where the N I/O-related counters include an I/O response time of the disk and a counter affecting the I/O response time; the I/O response time is a time between delivery of an operation request by an application and reception of a response of the disk to the request; and N is an integer greater than or equal to 2; determining, according to the N pieces of real-time data, whether the I/O response time is abnormal, where the I/O response time being abnormal indicates that the disk is incapable of normally running a service, and the I/O response time being normal indicates that the disk is capable of normally running a service; and outputting a detection result if the I/O response time is abnormal, where the detection result is used to represent that the I/O response time is abnormal.