Global Snapshots for Distributed Storage Rollback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern distributed storage systems face challenges in efficiently rolling back volume operations, which are resource-intensive and can block other operations, while existing snapshot technologies only allow for quick rollback of regular file operations, not volume operations.
Innovation Solution
A storage manager that creates global snapshots of distributed logical volumes, allowing for quick rollback of volume operations by decomposing them into regular operations on subvolumes, and includes features like snapshot management, reference counting, and resource allocation to facilitate efficient rollback and resource management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If volume operations are rolled back by performing opposite operations (e.g., removing a device to rollback adding), then rollback capability is provided, but resource intensity increases significantly due to re-distribution of objects
Solution Approach 1:
The system creates snapshots of the distributed logical volume at specific points in time before volume operations are performed. These snapshots capture the state of all subvolumes and their configurations. When a rollback is needed, the system simply restores from the pre-operation snapshot rather than performing complex opposite operations, thus avoiding the resource-intensive re-distribution of objects.
Solution Approach 2:
The invention creates copies of the volume state (snapshots) that include configuration information and subvolume data. Instead of modifying the actual volume during rollback operations, the system uses these copies to restore previous states efficiently, eliminating the need for resource-intensive data re-distribution that would occur with traditional opposite operations.
2Reliability
If volume operations are rolled back by performing opposite operations, then rollback is achieved, but other operations are blocked during the rollback process
Solution Approach 1:
Snapshots are created and stored before volume operations execute. This preliminary capture of state allows rollbacks to be performed by simply restoring from the snapshot without interfering with ongoing operations. The rollback process becomes a separate restoration operation rather than a blocking re-distribution process.
Solution Approach 2:
The volume is divided into subvolumes, and snapshots capture the state of each subvolume independently. This segmentation allows parallel processing where snapshot creation and volume operations can proceed simultaneously without blocking each other, maintaining high operational throughput while enabling reliable rollbacks.
3Reliability
If snapshots are created by copying all data of subvolumes, then complete backup is achieved, but storage space and time are consumed
Solution Approach 1:
The system extracts only the essential configuration information and metadata needed for rollback from the subvolumes, rather than copying entire data sets. The snapshots store configuration data structures that describe the volume state, which are much smaller than full data copies but sufficient for restoration purposes.
Solution Approach 2:
The snapshot implementation changes from storing complete data copies to storing configuration parameters and metadata that describe the volume state. This parameter-based approach maintains snapshot completeness for rollback purposes while dramatically reducing the storage space required.
4Reliability
If configuration information is stored to enable rollback, then rollback capability is provided, but system complexity increases
Solution Approach 1:
The configuration manager and snapshot manager are integrated into the existing storage management architecture, allowing the same components to handle both normal volume operations and rollback operations. This multi-functionality approach enables rollback capability without adding separate complex systems, as the existing management infrastructure is extended to handle snapshot and restoration tasks.
Data Source
AI summary
A storage manager for managing storage of files on a distributed logical volume includes a configuration manager configured to manage an archive of configurations of the distributed logical volume, and a snapshot manager configured to create a global snapshot of the distributed logical volume and, in response to a roll-back request, restore a previous global snapshot of the distributed logical volume based on a stored configuration of the distributed logical volume and stored subvolume snapshots of subvolumes of the distributed logical volume.