Intelligent Parallel Cloud Backup Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data recovery processes from cloud backups are time-consuming due to the limitations of Wide Area Network (WAN) speeds, resulting in significant downtime, especially when restoring large datasets like 10 Terabytes over a 1 Gb/s link which can take approximately 27 hours.
Innovation Solution
Implementing a data protection system that performs intelligent parallel recovery from multiple backup copies stored across different locations, optimizing data download by splitting data into segments and distributing them across various backup sites based on their download throughput, allowing simultaneous download from multiple locations to minimize overall recovery time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is recovered from a single cloud backup copy over WAN, then data protection is ensured, but recovery time becomes excessively long
Solution Approach 1:
The patent segments the backup copy into multiple data portions and distributes them across multiple cloud storage locations. During recovery, these segments are simultaneously downloaded from multiple sources in parallel, dramatically reducing total recovery time while maintaining data integrity through proper segment assembly
Solution Approach 2:
The patent combines multiple cloud storage locations into a unified recovery system, where data portions from different locations work together to reconstruct the complete dataset. This merging of resources transforms a single-point bottleneck into a multi-path recovery architecture
2Loss of time
If data is segmented and downloaded from multiple cloud locations in parallel, then recovery time is reduced, but system complexity increases
Solution Approach 1:
The patent introduces a recovery manager as an intermediary component that coordinates the complex multi-location recovery process. This mediator handles segment tracking, download coordination, and assembly orchestration, shielding users from the underlying complexity while enabling parallel recovery operations
Solution Approach 2:
The patent transitions from a single-dimension recovery approach (one backup location) to a multi-dimensional architecture by distributing data portions across multiple cloud locations. This dimensional expansion enables parallel operations without proportionally increasing user-facing complexity
3Productivity
If multiple backup copies are stored across different cloud locations, then recovery speed is improved, but storage costs increase
Solution Approach 1:
Instead of creating complete duplicate backups at multiple locations, the patent segments a single backup into portions distributed across multiple cloud storage locations. This segmentation strategy achieves multi-location recovery capability while using significantly less total storage capacity than full duplicates
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Data protection operations including recovery operations are disclosed. A recovery operation is performed by downloading the backup to be recovered from multiple identical copies of the backup. Based on factors such as throughput, an optimal amount of data can be downloaded from each of the multiple copies. The portions downloaded from the multiple copies are rebuild or reassembled once downloaded. The assembled backup is then presented to a target for recovery.