Virtual Machine Storage Peer Pair for Rapid Host Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Virtual machines cannot be quickly recovered when the host machine goes down, leading to service interruptions due to the lack of accessible storage.
Innovation Solution
Forming a storage peer pair between two host machines to backup read and write requests, allowing the virtual machine to be started on a backup host and synchronizing data, enabling quick migration and recovery without shared storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If read and write requests are stored locally on the host machine, then storage access speed is improved, but system reliability deteriorates when the host machine fails
Solution Approach 1:
The patent creates a copy of the virtual machine's storage data on a peer host machine. When the primary host fails, the backup host can provide the copied data to maintain service continuity. This resolves the contradiction by having local storage for speed while maintaining reliability through data replication on another machine.
Solution Approach 2:
The patent introduces a peer host machine as an intermediary that stores backup data and can mediate storage requests when the primary host is unavailable. This intermediary resolves the contradiction by providing an alternative storage access path that maintains both speed (through direct local storage) and reliability (through the backup intermediary).
2Reliability
If data is synchronized from second host machine to backup host machine, then reliability is improved, but synchronization time increases
Solution Approach 1:
The patent implements incremental synchronization where only changed data blocks are transferred between host machines rather than synchronizing entire datasets. This partial action approach maintains data reliability while significantly reducing synchronization time and network overhead.
Solution Approach 2:
The patent performs initial full data synchronization before the virtual machine needs to be recovered, so that the backup host is already prepared with data. When failure occurs, only minimal updates need to be transferred, reducing the actual recovery time while maintaining reliability through pre-synchronized data.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present application discloses a quick recovery method, device and system of a virtual machine. The virtual machine quick recovery method comprises: forming a first host machine and a second host machine into a storage peer pair, such that read and write requests of a virtual machine of the first host machine are backed up in the second host machine; starting, by a backup host machine, the virtual machine of the first host machine when the first host machine goes down or powers off; receiving, by the virtual machine started in the backup host machine, read and write requests transmitted by a calling party, and judging whether full data synchronization from the second host machine to the backup host machine is finished or not; if the full data synchronization from the second host machine to the backup host machine is not finished, transmitting read and write requests to the second host machine; receiving, by the second host machine, the read and write requests, storing data to be written in the second host machine based on the write request, or acquiring specified data based on the read request and transmitting the specified data to the backup host machine; transmitting, by the backup host machine, the specified data to the calling party.