VP Extent Statistics for CDP Storage Restoration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage systems struggle to efficiently restore data and performance levels at a backup site to a specific point in time, particularly in tiered storage systems, where data is automatically moved between storage tiers based on IO activity, making it challenging to maintain optimal performance and data integrity.
Innovation Solution
A method and apparatus that restore a mirrored storage system's backup site to a selected point in time by determining the data state, writing data indicative of that state to the backup site storage array, and moving data extents to selected tiers to achieve a predetermined level of performance, utilizing journaling of statistical metadata to record IO activity and restore tiering and performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is automatically moved between storage tiers based on IO activity at the backup site, then storage performance is optimized, but the restored data state does not match the historical performance levels
Solution Approach 1:
The system captures and stores IO activity statistics (metadata) at the production site before failure occurs. During restoration, this pre-captured metadata is applied to the backup site data, preliminarily establishing the correct tiering configuration that reflects historical performance levels, thus avoiding the need to relearn IO patterns after restoration
Solution Approach 2:
The system copies IO activity statistics from the production site to the backup site. By replicating the metadata that tracks IO patterns, the backup site receives not only data copies but also the performance characteristics information, enabling it to reconstruct its tiering structure to match the production site's historical state
2Reliability
If IO activity statistics are captured and stored for each extent, then performance can be restored to historical levels, but system complexity increases
Solution Approach 1:
The system introduces metadata as an intermediary layer between the actual data and the tiering decision-making process. This metadata layer captures IO activity statistics without modifying the core data storage operations, simplifying the complexity by separating performance tracking from data management while enabling accurate performance restoration
Solution Approach 2:
The system changes the parameter being monitored from raw IO activity to aggregated statistics about IO patterns. By tracking statistical metadata (such as read/write frequencies, access patterns) rather than individual IO events, the system achieves accurate performance restoration with reduced processing overhead and system complexity
3Productivity
If the backup site mirrors the production site's tiering structure, then performance levels are maintained, but storage cost increases
Solution Approach 1:
The system applies local quality by using IO activity statistics to determine which specific data extents should be placed on which storage tiers at the backup site. Rather than uniformly maintaining high-performance tiering across all data, the system selectively tiers data based on its actual access patterns, optimizing performance for frequently accessed data while using lower-cost storage for inactive data
Data Source
AI summary
A replica site is restored to a selected point in time by determining data state at the selected point in time, writing data indicative of that data state to the replica site storage array, and moving extents of the data written to the replica site storage array to selected tiers in order to achieve a predetermined level of performance. A journal of statistical meta data indicative of IO activity may be used to select the tiers.