Database Snapshot Recovery via Recoverable Chains
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data management systems face challenges in efficiently recovering databases and applications with reduced downtime, particularly in virtualized infrastructure environments where high workload consolidation and instant granular recovery are required.
Innovation Solution
The implementation of an integrated data management and storage system that captures and stores snapshots of databases at frequent intervals, along with database transaction logs, to create recoverable chains. These chains allow for near point-in-time recovery by combining snapshots and transaction logs, thereby reducing recovery time objectives (RTOs).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all stored snapshots and logs are enumerated during recovery, then complete database recovery is achieved, but recovery time increases
Solution Approach 1:
The patent extracts and removes unnecessary snapshots and logs from the recovery process by implementing expunge operations that eliminate superseded or redundant recovery points. This allows the system to achieve complete recovery using only the essential snapshots and logs in the recoverable chain, rather than enumerating all stored snapshots and logs, thereby reducing recovery time while maintaining recovery completeness
Solution Approach 2:
The system performs preliminary organization of snapshots and logs into structured recoverable chains during normal operation, maintaining metadata that identifies valid recovery sequences. This preliminary structuring enables the recovery process to directly access the necessary snapshots and logs without needing to enumerate or search through all stored snapshots and logs, reducing recovery time while ensuring complete recovery
2Measurement precision
If snapshot frequency is increased to reduce recovery point gap, then near point-in-time recovery is achieved, but storage requirements and system overhead increase
Solution Approach 1:
The patent applies different retention and storage policies to different snapshots based on their position in the recoverable chain and their utility for recovery. Recently taken snapshots that are part of active recoverable chains are retained with higher priority, while older or superseded snapshots are eligible for expungement. This local differentiation allows near point-in-time recovery precision while optimizing storage resource usage by not uniformly retaining all snapshots at high frequency
Solution Approach 2:
The system implements a mechanism to discard superseded or redundant snapshots through expunge operations while maintaining the ability to recover any needed point in time through the preserved recoverable chains. This selective discarding removes unnecessary storage burden from high-frequency snapshotting while preserving recovery precision for critical time points
3Loss of time
If recoverable chains are implemented to reduce recovery time, then rapid recovery is achieved, but system complexity increases
Solution Approach 1:
The patent introduces an intermediary expunge operation that acts as a mediator between the snapshot creation process and the recovery process. This expunge mechanism automatically manages the lifecycle of snapshots and logs, identifying and removing superseded recovery points while maintaining valid recoverable chains. This intermediary layer simplifies the overall system by automating the complex task of managing recoverable chains, reducing manual intervention requirements while enabling rapid recovery
Data Source
AI summary
Methods and systems for recovering databases and other applications with reduced downtime using recoverable chains are described. When protecting or backing up a database, an integrated data management and storage system may capture snapshots of the database at a snapshot frequency (e.g., at 15 minute intervals) and acquire database transaction logs that include data changes of the database at a frequency that is greater than the snapshot frequency. The integrated data management and storage system may generate a particular point in time version of the database by identifying a recoverable chain that comprises a combination of one or more database snapshots including the snapshot closest to the recovery point and one or more database transaction logs.