Virtual Machine Data Storage Recovery via Metadata Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing data recovery processes for virtual machines on virtualization platforms are inefficient, leading to potential data loss due to the long IO stack and susceptibility to failures at any level during recovery.
Innovation Solution
A data storage method for virtual machines that involves acquiring a starting instruction, loading a storage device, accessing and verifying metadata information, determining data storage states based on data region bitmap information, and sending target data for storage only when both files are in a normal state.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional storage recovery methods are used, then data recovery can be achieved, but the IO stack is very long and any failure at any level leads to data loss
Solution Approach 1:
The patent segments the data recovery process into independent file-level operations with parallel processing. Each file is treated as an independent unit with its own metadata verification and data region bitmap checking, allowing simultaneous recovery operations across multiple files without blocking each other, thus reducing the overall IO stack depth and failure propagation risk.
Solution Approach 2:
The patent performs preliminary verification of metadata information and data region bitmap before actual data recovery. By checking the integrity and status of file metadata and bitmap structures in advance, the system identifies and skips corrupted files, preventing recovery failures from propagating through the entire IO stack and ensuring more reliable data recovery.
2Quantity of substance
If comprehensive data recovery is performed, then data completeness is improved, but recovery time increases due to lengthy IO stack
Solution Approach 1:
The patent implements partial recovery by selectively processing only files that pass metadata verification and bitmap integrity checks. Instead of attempting to recover all files sequentially through a deep IO stack, the system identifies valid files and recovers only those, skipping corrupted or inaccessible files, thus reducing recovery time while maintaining completeness for recoverable data.
Solution Approach 2:
The patent enables continuous parallel recovery operations across multiple files simultaneously. By maintaining independent recovery threads for each file and using bitmap information to track recovery status, the system continues useful recovery actions without interruption from IO stack blocking, significantly reducing total recovery time while preserving data completeness for all recoverable files.
3Reliability
If file-level verification is implemented, then data storage reliability is improved, but verification complexity increases
Solution Approach 1:
The patent introduces data region bitmap information as an intermediary structure between file metadata and actual data regions. The bitmap serves as a mediator that translates complex file-level verification requirements into simple bit-checking operations, making the verification process less complex while maintaining high reliability through comprehensive file status tracking.
Data Source
AI summary
Disclosed are a data storage method and apparatus of a virtual machine, a virtual machine, and a storage medium. The method includes: acquiring a starting instruction of a virtual machine, and loading a storage device on the basis of the starting instruction; accessing metadata information recorded in first storage regions of at least two corresponding files in the storage device, and verifying the metadata information to obtain a verification result; when the verification result is normal, accessing data region bitmap information recorded in second storage regions of the at least two files; determining, on the basis of the data region bitmap information, data storage states of the at least two files; and when the data storage states of the at least two files are both normal, sending target data to the at least two files to perform data storage.