Virtual Machine High Availability via Shared Storage During Network Isolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In computer virtualization, when a server loses network connectivity and becomes isolated, existing solutions either shut down or power off virtual machines, resulting in loss of their execution state, which disrupts availability and loses guest operating system and application state.
Innovation Solution
A method that detects network isolation of a host and transfers the execution of a virtual machine to another host using a shared storage system, maintaining the execution state of the virtual machine through live migration, suspend-resume, or snapshot reversion operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If virtual machines are shut down or powered off when network isolation is detected, then the system responds to the failure condition, but the execution state is lost and availability is disrupted
Solution Approach 1:
The system performs preliminary actions by detecting network isolation conditions and initiating VM migration before complete system failure occurs. The isolation detection mechanism triggers pre-configured migration procedures, allowing VMs to be moved to alternative hosts while maintaining their execution state, thus avoiding downtime and state loss that would occur with traditional shutdown approaches
Solution Approach 2:
The patent introduces an intermediary mechanism (isolation detection and migration orchestration system) that mediates between the network failure condition and the VM execution state. This intermediary layer detects isolation events and coordinates the migration process through shared storage, enabling seamless VM relocation without direct interruption to the execution state, thereby resolving the contradiction between responding to failure and maintaining availability
2Reliability
If virtual machines are restarted on other servers after network isolation, then availability is restored, but significant disruption occurs and execution state is lost
Solution Approach 1:
The system creates and maintains copies of VM execution states through shared storage infrastructure. When network isolation occurs, the VM state is already replicated on the shared storage system, allowing immediate restoration on alternative hosts by retrieving the copied state rather than performing cold restarts. This copying mechanism eliminates service disruption and preserves execution state while restoring availability
Solution Approach 2:
The migration infrastructure performs preliminary setup by establishing shared storage connectivity and pre-configuring migration capabilities before isolation events occur. This preliminary preparation enables rapid state transfer and minimizes disruption when restoration is needed, allowing the system to quickly recover VM execution on alternative hosts without significant productivity loss
3Reliability
If the isolated server shuts down virtual machines, then the failure condition is addressed, but guest operating system and application state is lost
Solution Approach 1:
The shared storage system serves as an intermediary that preserves VM execution states during isolation events. Instead of the isolated server directly shutting down VMs and losing state, the intermediary storage infrastructure maintains the state data, allowing recovery of guest OS and application state on alternative hosts. This intermediary mechanism decouples the failure condition from information loss
Solution Approach 2:
The patent extracts the VM execution state from the isolated server's local memory and storage, separating it from the failed host system. By taking out the critical execution state data and placing it in the shared storage system, the invention prevents loss of guest OS and application state even when the original server must be taken offline to address the failure condition
Data Source
AI summary
In case of network isolation of a host executing one or more virtual machines, the state of the one of more virtual machines is saved using a variety of isolation response mechanisms. Isolation responses may include a live migration to another host using a shared storage system connected to both hosts, a virtual machine suspend and resume operation, and a snapshot reversion operation. The execution state of the virtual machine(s) running on the isolated host, which includes the state of the guest operating system and any running applications, are maintained in the other host, even after host isolation has occurred.