Log-Structured Filesystem Recovery via Cloud Replica Sorting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Log-structured filesystems face challenges in recovering from disasters due to out-of-order writes in replicas, which break the temporal ordering needed for proper operation, leading to inconsistencies and corruption of filesystem data and metadata.
Innovation Solution
The solution involves using a cloud-based object store as target storage for physical replicas, ensuring temporal order preservation by mirroring data and metadata across block and object storage, and implementing methods to recover filesystem data and metadata from physical replicas, including redundant metadata storage and logical sorting of objects to maintain consistency and integrity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physical replication is used for disaster recovery, then data availability is improved, but temporal ordering is lost leading to out-of-order writes
Solution Approach 1:
The patent introduces a mediator layer (the recovery system with sorting logic) between the physical replica storage and the log-structured filesystem recovery process. This mediator sorts the retrieved data blocks according to their logical addresses before presenting them to the filesystem, thereby restoring temporal ordering without sacrificing the availability benefits of physical replication
Solution Approach 2:
The patent performs preliminary sorting and reordering of data blocks during the recovery process before the filesystem becomes operational. By pre-organizing the replicated data in the correct temporal sequence, the system ensures that when the filesystem mounts, it receives properly ordered data without needing to maintain ordering during the replication process itself
2Productivity
If out-of-order writes occur in replicas, then recovery speed is improved, but data consistency is lost leading to corruption
Solution Approach 1:
The recovery system acts as an intermediary that receives out-of-order data blocks from the replica storage, sorts them based on logical address information, and outputs them in the correct sequential order. This mediator approach maintains recovery speed by allowing parallel retrieval while ensuring data consistency through systematic reordering
Solution Approach 2:
The patent segments the recovery process into distinct phases: data retrieval from replica, sorting/reordering based on logical addresses, and presentation to the filesystem. This segmentation allows the system to handle out-of-order writes efficiently by processing retrieval in parallel while maintaining consistency through the intermediate sorting stage
Data Source
AI summary
Systems, methods and apparatuses can present recovery of a log-structured file system. Embodiments can provide sending a request to a cloud store for backup data. The set of backup data received from the cloud store can comprise a set of data and metadata objects. The set of metadata objects can be read in a logical order. Each metadata object can be written from the set of data and metadata objects into block storage of the log-structured file system.