SSD Data Prediction Model for Dynamic Storage Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing SSD storage technologies have high storage costs and affect the usage lifetime due to inefficient data management and static classification methods.
Innovation Solution
A data storing method and apparatus that predicts reading and writing data information using a data prediction model, dynamically determining physical storage areas based on historical data patterns to optimize storage allocation and adapt to changing workloads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If static classification methods are used for cold and hot data, then storage management is simple, but storage cost is high and usage lifetime is reduced
Solution Approach 1:
The patent implements dynamic data classification by continuously monitoring I/O workload characteristics and automatically adjusting data placement policies. The system transitions from static cold-hot data classification to dynamic classification based on real-time workload analysis, where data is classified into different categories (sequential, random, mixed, etc.) based on current access patterns. This dynamic approach optimizes wear distribution and extends SSD lifetime while maintaining adaptive storage management.
Solution Approach 2:
The system employs feedback mechanisms by continuously collecting I/O workload information, analyzing access patterns, and using this feedback to adjust data placement decisions. The controller monitors read/write operations, identifies workload characteristics, and dynamically modifies classification policies based on observed patterns. This closed-loop feedback system enables adaptive optimization of storage performance and reliability without requiring complex manual configuration.
2Ease of operation
If manual customization of stream addresses is performed, then data placement can be controlled, but adaptability to changing workloads is lost
Solution Approach 1:
The system implements self-service by automatically analyzing I/O workload characteristics and making intelligent data placement decisions without requiring manual intervention. The controller autonomously monitors access patterns, classifies data based on workload types (sequential, random, mixed, small-file, large-file), and dynamically adjusts stream address assignments. This self-service capability provides both ease of operation through automation and adaptability to changing workloads, eliminating the need for manual policy customization while maintaining optimal data placement.
Solution Approach 2:
The patent enables dynamic stream address assignment by continuously adapting data placement policies based on real-time workload analysis. Instead of fixed manual assignments, the system dynamically adjusts stream addresses according to observed access patterns and workload characteristics. This dynamic approach maintains ease of operation through automated control while achieving high adaptability to varying workload conditions.
3Device complexity
If simple cold-hot data classification is used, then storage management is straightforward, but storage performance optimization is limited
Solution Approach 1:
The patent segments data classification into multiple detailed categories beyond simple cold-hot division. Data is classified into specific types including sequential access, random access, mixed access, small-file operations, large-file operations, and other patterns. This fine-grained segmentation enables more precise optimization of storage performance for different workload types while maintaining manageable complexity through automated analysis and classification.
Solution Approach 2:
The system dynamically changes classification parameters based on observed workload characteristics. Instead of using fixed classification thresholds, the system adjusts classification criteria according to real-time I/O patterns, access frequencies, and workload types. This parameter adaptation enables enhanced storage performance optimization while keeping the classification system manageable through automated parameter tuning.
Data Source
AI summary
Provided is a method including acquiring reading and writing information of logical chunks of a storage apparatus in a number of historical periods before a current time, and predicting reading and writing information of the logical chunks in a next period according to reading and writing information of the logical chunks in a number of historical periods before the current time and a data prediction model. The data prediction model indicates a relationship between reading and writing information of the logical chunks in a next period and reading and writing information of the logical chunks in a number of historical periods before the current time.