Geographically Distributed Tree Recovery via Peer Zone Journaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-based storage systems face significant data loss due to unavailability of tree chunks, which can result from corruption or hardware failures, despite data protection techniques like mirroring and erasure coding, leading to incomplete or lost trees.
Innovation Solution
A geographically distributed storage environment is used to recover impacted trees by sending update-related information from an impacted zone to a peer zone, identifying a recovery range, and processing a tree recovery journal to regenerate the impacted tree part, ensuring non-disruptive recovery without adverse effects on the storage system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data protection techniques such as mirroring and erasure coding are used, then data reliability is improved, but data loss still occurs when tree chunks become unavailable due to corruption or hardware failures
Solution Approach 1:
The patent creates a peer tree in a peer zone that serves as a copy of the original tree in the impacted zone. When tree chunks become unavailable in the impacted zone, the peer tree can be used to recover the lost data by copying the necessary tree elements from the peer zone back to the impacted zone, thus resolving the data loss issue while maintaining the reliability improvements from mirroring and erasure coding
Solution Approach 2:
The peer tree is proactively created and maintained in the peer zone before any data loss occurs. This preliminary action ensures that when tree chunks become unavailable, the recovery process can immediately utilize the pre-existing peer tree without needing to reconstruct it from scratch, reducing recovery time and ensuring continuous data availability
2Reliability
If tree chunks are stored in a geographically distributed environment, then system availability is improved, but complexity of data recovery increases
Solution Approach 1:
The patent divides the recovery process into distinct segments: identifying the impacted tree, determining the recovery range, selecting a peer zone, and transferring the peer tree. This segmentation of the recovery process into manageable steps reduces the overall complexity by making each step independent and well-defined, while maintaining the benefits of geographic distribution for system availability
Solution Approach 2:
The patent introduces a peer zone as an intermediary that holds a peer tree serving as a backup. This intermediary structure simplifies the recovery process by providing a straightforward mechanism to retrieve lost tree elements without requiring complex cross-zone coordination or reconstruction algorithms, thus reducing recovery complexity while maintaining geographic distribution benefits
Data Source
AI summary
The described technology is generally directed towards recovery of an impacted (damaged) tree in an impacted zone in a geographically distributed data storage environment, using a peer tree in a remote zone. A peer zone is selected for recovery, and updated to have current data. Logic at the impacted zone requests recovery of an impacted tree, identifying one or more recovery ranges for which recovery data is needed. Logic at the peer zone locates missing objects via the peer tree, and provides recovery information to the impacted zone, by which recovery of the impacted tree is able to be accomplished. For example, a replication journal may be returned as a tree recovery journal comprising add leaf (object) instructions, whereby the impacted zone processes the journal with respect to a reduced representation of the impacted tree to obtain a recovered tree.