Restore Point Selection for Continuous Data Protection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In continuous data protection environments, the large number of restore points and event logs leads to inefficient recovery processes, resulting in prolonged recovery times due to the need for manual testing and selection of valid restore points, especially in cases of data corruption or loss.

Innovation Solution

A computer-implemented method for efficiently selecting restore points by identifying interesting restore points associated with unusual I/O activity, assigning confidence scores, grouping them for recovery scheduling, and validating their data validity using machine learning models and validation functions across available nodes, allowing for parallel processing and resource optimization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If all restore points are manually tested and selected for recovery, then data validity can be ensured, but recovery time is significantly prolonged

Engineering Contradiction:

Improvedata validityVSAvoidrecovery time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system automatically identifies, scores, and validates restore points without manual intervention. The machine learning model autonomously evaluates restore point usefulness based on I/O activity patterns, and the system automatically tests and selects valid restore points, eliminating the need for manual testing while ensuring data validity.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

Manual testing and selection processes are replaced with automated machine learning models and validation functions. The ML model substitutes human judgment in identifying useful restore points, and automated validation functions replace manual data verification, dramatically reducing recovery time while maintaining reliability.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If the system maintains a continuous log of all data changes, then complete recovery capability is achieved, but the log size becomes unmanageably large

Engineering Contradiction:

Improverecovery capabilityVSAvoidlog size

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The system extracts only the most useful restore points from the continuous log based on I/O activity analysis. By identifying and isolating restore points that are most likely to be needed for recovery (those associated with unusual I/O activity), the system reduces the effective log size that must be managed and processed, while maintaining complete recovery capability for critical points.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

Different portions of the log are treated differently based on their characteristics. Restore points associated with unusual I/O activity are identified as having higher value and are prioritized for retention and faster access. The system applies different quality standards to different log entries, focusing resources on the most valuable portions of the log.

Inventive Principle:

Principle #3Local quality

3Productivity

If multiple restore points are validated in parallel across available nodes, then validation speed increases, but system complexity increases

Engineering Contradiction:

Improvevalidation speedVSAvoidsystem complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The validation process is segmented into independent tasks that can be executed in parallel across multiple nodes. Each restore point validation is divided into discrete, manageable units that can be assigned to different computing nodes simultaneously. This segmentation enables parallel processing while keeping individual task complexity manageable.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The validation function is designed as a universal, reusable component that can operate on any restore point across any node. The same validation logic and machine learning model are applied consistently across all parallel operations, reducing the complexity that would arise from having node-specific or restore-point-specific validation routines.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS11775399B1Efficient recovery in continuous data protection environments

Publication Date: 2023.10.03 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

A computer-implemented method, a computer system and a computer program product efficiently select restore points in a continuous data protection environment. The method includes receiving log entries that include restore points that correspond to data stored on nodes in the continuous data protection environment. The method also includes identifying an interesting restore point from the log entries. The method further includes grouping the interesting restore point for recovery based on one or more of a confidence score and a restore time. In addition, the method includes loading the group of interesting restore points on available nodes in the continuous data protection environment. The method also includes determining whether the data corresponding to each interesting restore point in the group is valid using a validation function. Lastly, the method includes discarding the interesting restore point when the data corresponding to the interesting restore point is not valid.