Virtual Machine Recovery with Prioritized Block Prefetching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current backup systems face challenges in efficiently copying large data volumes while an operating system is actively writing to the source volume, leading to performance issues and increased storage requirements due to the need for copy-on-write mode, which can result in prolonged copying processes and resource bottlenecks.
Innovation Solution
The system implements a method to copy a source volume to a target volume by designating blocks, identifying written blocks, and entering copy-on-write mode only when necessary, reducing the time spent in copy-on-write mode and optimizing data transfer through predictive and lookahead optimizations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If copy-on-write mode is used to copy the source volume while the operating system is actively writing, then data consistency is maintained, but the copying process time is prolonged and system performance is impacted
Solution Approach 1:
The patent applies partial action by selectively entering copy-on-write mode only for specific blocks that require consistency guarantees, rather than forcing the entire copying process into copy-on-write mode. This allows the system to maintain data consistency for critical blocks while avoiding the performance penalty across the entire volume, thus reducing overall copying time while preserving reliability where needed.
Solution Approach 2:
The system dynamically transitions between copy-on-write mode and normal copying mode based on real-time conditions. The copy process can switch modes for different blocks or time periods, allowing optimal performance characteristics to be maintained throughout the copying process while ensuring consistency when required by system state or block importance.
2Reliability
If copy-on-write mode is entered during the copying process, then data integrity is preserved, but storage requirements increase and resource bottlenecks occur
Solution Approach 1:
The patent applies local quality by applying copy-on-write protection selectively to specific blocks or regions of the volume rather than uniformly across the entire source volume. This allows data integrity to be maintained for blocks that require it while avoiding the storage overhead and resource consumption associated with copy-on-write for blocks that do not require such guarantees, thus reducing overall storage requirements.
3Reliability
If the entire source volume is copied using traditional methods, then complete backup is achieved, but system performance is significantly impacted and resource bottlenecks occur
Solution Approach 1:
The patent segments the source volume into multiple blocks and processes them independently, allowing selective application of copy-on-write mode to only those blocks that require consistency guarantees. This segmentation enables the backup process to maintain completeness while minimizing performance impact by avoiding unnecessary copy-on-write operations on blocks that can be copied using more efficient methods.
Solution Approach 2:
The copying system is designed to support multiple copying modes (copy-on-write and normal copying) within a single unified process, allowing it to adaptively select the appropriate mode for each block based on system conditions, block importance, and consistency requirements. This multi-functionality enables complete backup while maintaining system performance by avoiding uniform application of the more restrictive copy-on-write mode.
Data Source
AI summary
The Prioritization and Source-Nonspecific Based Virtual Machine Recovery Apparatuses, Methods and Systems (“MBR”) transforms backup configuration request, restore request inputs via MBR components into backup configuration response, restore response outputs. A restore request to build a recovery virtual machine using a backup disk image of a source device stored on a backup device is obtained. A virtual machine recovery environment comprising a guest virtual machine running a guest hypervisor is booted. A recovery virtual machine configuration identifying source-nonspecific software is determined. A recovery prioritization index for data blocks of the backup disk image is determined. Essential data blocks of the backup disk image are prefetched. An operating system of a nested guest abridged virtual machine is booted. Command data blocks are fetched when a user command is received. Remaining data blocks are fetched in priority order. The nested guest complete virtual machine is migrated to a non-nested guest virtual machine.