Deep Learning Model for Predicting RPO Drifts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current disaster recovery (DR) operations fail to provide early warnings for workload/application recovery point objective (RPO) related issues, leading to missed business service level agreements (SLAs), penalties, and downtime, with manual root cause analysis being time-consuming and inefficient, and lacking predictive capabilities for future RPO values.
Innovation Solution
A computer-implemented method using deep learning to predict RPO drifts for a definite future time window with a confidence score, extracting influencing factors using interpretable AI, and enabling continual learning by ingesting ground truth, allowing for proactive SLA achievement and resiliency management across multiple domains.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual root cause analysis is used for RPO deviations, then detailed investigation can be performed, but the process is time-consuming and delays response time
Solution Approach 1:
The system performs preliminary actions by continuously collecting and preprocessing RPO metrics, system configuration changes, performance data, and event logs before deviations occur. This pre-positioning of data and computational frameworks enables immediate analysis when deviations happen, eliminating the need for time-consuming manual data gathering and initial assessment.
Solution Approach 2:
An AI-based intermediary system is introduced between the complex disaster recovery infrastructure and the analysis process. This intermediary automatically correlates RPO metrics with system configuration changes, performance data, and event logs, providing structured insights that bridge the gap between raw data and root cause identification, thereby reducing manual investigation time while maintaining accuracy.
2Reliability
If current RPO values are calculated in real-time, then monitoring is provided, but no information is available about future values for proactive management
Solution Approach 1:
The system performs preliminary analysis by continuously processing historical RPO metrics, system configuration changes, and performance data to establish baseline patterns and relationships. This pre-computation of correlations and trends enables the system to generate predictive insights about future RPO values before actual deviations occur, allowing proactive intervention.
Solution Approach 2:
A feedback mechanism is implemented where the system continuously monitors actual RPO values, compares them with predicted values, and uses the discrepancies to refine its predictive models. This closed-loop feedback ensures that the system learns from past performance and improves its ability to predict future RPO drifts, enhancing both monitoring reliability and predictive accuracy over time.
3Adaptability or versatility
If deep learning models are trained on multiple domains, then generalization capability is improved, but model complexity and training requirements increase
Solution Approach 1:
A universal deep learning model architecture is designed that can process and analyze multiple types of data (RPO metrics, system configuration changes, performance data, event logs) across different disaster recovery domains. The model uses shared layers for common feature extraction and domain-specific layers for specialized analysis, enabling it to generalize across domains while managing complexity through modular design and transfer learning techniques.
Data Source
AI summary
A computer-implemented method includes: generating, by a computing device, a multidimensional hyperspace encompassing a plurality of features based on input data; generating, by the computing device, a plurality of sequential arrays of a fixed length based on the input data; generating, by the computing device, a final sequential array of a predetermined shape based on the plurality of sequential arrays of the fixed length; generating, by the computing device, a training data set for a deep learning model based on the final sequential array of the predetermined shape and the multidimensional hyperspace; and training, by the computing device, a deep learning model using the generated training data set.