Data Storage Optimizer Cost Metrics for Heterogeneous Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage systems face challenges in optimizing data movement between devices with different performance characteristics to enhance system performance, as current optimizers only consider devices with similar performance characteristics for load balancing, neglecting potential improvements from moving data between devices with varying capabilities.
Innovation Solution
A method is introduced to determine whether to move data between devices by calculating cost metrics based on response times and processing rates, adjusting these values according to performance characteristics and RAID protection levels, and evaluating alternative configurations to identify improvements in data storage system performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data movement is optimized only between devices with similar performance characteristics, then load balancing is achieved, but system performance improvement is limited
Solution Approach 1:
The patent changes the evaluation parameters from simple utilization metrics to cost metrics that incorporate response time and processing rate. This allows the optimizer to evaluate data movement between devices with different performance characteristics by calculating the impact on overall system cost, enabling adaptation to heterogeneous device environments while maintaining performance optimization.
Solution Approach 2:
The patent introduces dynamic evaluation of data movement decisions based on real-time or near-real-time performance characteristics. The cost metric calculation adapts to current device states (response time, processing rate) rather than static device classifications, allowing the system to dynamically determine optimal data placement across devices with varying performance levels.
2Productivity
If data is moved between devices with different performance characteristics, then potential performance improvements are achieved, but complexity of evaluation increases
Solution Approach 1:
The patent simplifies the evaluation complexity by changing from multi-dimensional device comparison to a unified cost metric calculation. By aggregating response time and processing rate into a single cost value for each device, the system can compare data movement options across heterogeneous devices using a consistent metric, reducing the complexity of evaluating cross-device performance interactions.
Solution Approach 2:
The cost metric serves as an intermediary that mediates between different device performance characteristics. Instead of directly comparing devices with different response times and processing rates, the patent uses the cost metric as an intermediate representation that captures the combined effect of these parameters, simplifying the optimization decision process.
3Measurement precision
If response time and processing rate are considered in cost metric calculation, then accurate performance evaluation is achieved, but computational overhead increases
Solution Approach 1:
The patent changes the evaluation from considering multiple independent device parameters to calculating a derived cost metric that combines response time and processing rate. This parameter transformation reduces the computational burden by pre-calculating or periodically updating the cost metric rather than performing complex multi-parameter analysis for each data movement evaluation, thereby reducing computational energy while maintaining evaluation accuracy.
Data Source
AI summary
Described are techniques for determining whether to move data between devices. A first cost metric is determined for a first device and a second device. At least one portion of data is selected for movement from one of the first device and the second device to another of the first device and the second device. A second cost metric is determined for the first device and the second device based on moving said at least one portion of data between said first device and said second device. It is determined using the first cost metric and the second cost metric whether to move the at least one portion of data.