VM Recovery via Multi-Source State Detection and Priority Relocation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing virtual machine high availability solutions in distributed environments fail to accurately determine and prioritize the recovery of critical virtual machines, leading to resource wastage and potential erroneous fault detection.

Innovation Solution

A method involving independent computing assemblies on each physical machine reporting states to a database, with a highly available controller checking and processing abnormalities by pinging and monitoring network connectivity, and executing relocation operations based on secondary voting to ensure critical virtual machines are recovered preferentially.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If all virtual machines in a highly available group are treated equally for recovery, then the high availability mechanism is simple to implement, but critical virtual machines cannot be prioritized for recovery

Engineering Contradiction:
Improverecovery priority of critical virtual machinesVSAvoidcomplexity of high availability mechanism
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces priority levels for different virtual machines within the highly available group. Critical virtual machines are assigned higher priorities while non-critical ones have lower priorities. This local differentiation allows the system to treat different virtual machines differently based on their importance, enabling preferential recovery of critical VMs without completely redesigning the high availability mechanism.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If only the control node detects physical machine states through heartbeating, then the detection mechanism is simple, but the detection accuracy is insufficient and may cause erroneous determinations

Engineering Contradiction:
Improveaccuracy of physical machine state detectionVSAvoidcomplexity of detection mechanism
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent combines multiple detection approaches: the control node's heartbeat detection, peer-to-peer detection among physical machines, and detection by virtual machines themselves. These multiple detection sources are merged to form a comprehensive state assessment. The system cross-validates detection results from different sources to reduce erroneous determinations while maintaining reasonable system complexity.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent implements a feedback mechanism where detection results from multiple sources are continuously monitored and used to adjust the overall state determination. When detection results conflict or are inconsistent, the system uses feedback loops to re-evaluate and resolve the true state of physical machines, improving detection accuracy through iterative validation.

Inventive Principle:
Principle #23Feedback

3Reliability

If virtual machines are relocated upon detecting any physical machine abnormality, then the high availability is maintained, but resources are wasted when the physical machine is actually functioning normally

Engineering Contradiction:
Improvehigh availability assuranceVSAvoidresource wastage from unnecessary relocation
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The patent implements preliminary verification steps before triggering virtual machine relocation. When an abnormality is detected, the system first performs additional validation checks to confirm the true state of the physical machine. Only after confirming that the physical machine is indeed abnormal does the system proceed with relocation. This preliminary action prevents unnecessary resource consumption from false positive detections.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3142011B9Anomaly recovery method for virtual machine in distributed environment
Publication Date: 2019.05.29 CHINA UNIONPAY
  • EP3142011B9 patent drawingFigure 1

AI summary

The invention proposes a virtual machine abnormity recovering method in distributed environment, comprising: running an independent computing assembly on each physical machine on which a virtual machine resides, wherein the computing assembly periodically reports the current running state of the corresponding physical machine to a state database; periodically polling the state database by a highly available controller so as to check the running state of all the physical machines in a physical machine group under the control of the highly available controller; and executing subsequent abnormity processing operation if the running state of only one physical machine in the physical machine group is abnormal so as to ensure that virtual machines on the physical machine whose running state is abnormal continues running normally. The virtual machine abnormity recovering method in distributed environment disclosed by the invention can accurately determine and efficiently handle faults of the physical machines in a distributed environment.