Predictive Object Tiering via Heat Value Modification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current tiered data storage systems in cloud environments face inefficiencies in dynamically adjusting the storage tiers of objects based on their metadata, leading to suboptimal performance and latency in data access.
Innovation Solution
A method for predictive object tiering that identifies objects and their associated metadata, modifying heat values of related objects in response to access, to dynamically adjust their storage tiers and optimize data access latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If objects are stored in a tiered storage system based on current access patterns, then storage optimization is achieved, but the system cannot proactively prepare for future access patterns leading to suboptimal performance
Solution Approach 1:
The system performs preliminary actions by predicting future access patterns before they occur. It uses machine learning models to analyze current access patterns and metadata relationships, then proactively moves objects to appropriate storage tiers in advance, preparing the storage system for anticipated access demands and reducing future latency.
Solution Approach 2:
The system implements feedback loops where access patterns and metadata relationships are continuously monitored and fed into machine learning models. These models learn from historical data and refine their predictions, creating a feedback-driven tiering strategy that adapts to changing access patterns over time.
2Productivity
If dynamic tiering adjustments are made based on real-time access patterns, then storage optimization improves, but system complexity increases
Solution Approach 1:
The system performs self-service by automatically analyzing its own access patterns and metadata relationships without external intervention. Machine learning models process internal data to generate tiering decisions autonomously, reducing the need for complex external control systems and manual configuration.
Solution Approach 2:
The system changes parameters such as heat values and metadata relationships dynamically to drive tiering decisions. By transforming access patterns into quantitative parameters that feed machine learning models, the system achieves adaptive optimization through parameter transformation rather than complex rule-based logic.
3Adaptability or versatility
If heat values are modified based on metadata relationships, then predictive tiering is enabled, but measurement and detection difficulty increases
Solution Approach 1:
The system replaces mechanical rule-based detection with machine learning-based detection. Instead of using explicit rules to identify metadata relationships, the system uses trained models that automatically detect and learn relationships from data, substituting complex manual detection mechanisms with automated intelligent systems.
Data Source
AI summary
In one general embodiment, a computer-implemented method is provided. The computer-implemented method includes identifying a first object in a tiered storage system, and identifying first metadata associated with the first object. The computer-implemented method also includes identifying a second object utilizing the first metadata of the first object. The second object is associated with second metadata. Moreover, the computer-implemented method includes, in response to an access of the first object, modifying a heat value associated with the second object.