Storage Pool Load Balancing via Processing Power Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional mapped RAID systems fail to effectively utilize storage resources across multiple storage pools, leading to inefficient data distribution and longer response times due to mismatched processing power and access frequencies among storage devices.
Innovation Solution
A method that determines the processing power of storage devices within each storage pool, divides them into sets based on similar processing power, and redistributes data based on workload to balance I/O loads across pools, ensuring hot data is stored on high-processing devices and cold data is optimized for lower-processing devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If extents of mapped RAID are distributed over multiple storage pools with different device types, then storage resource utilization is improved, but data distribution efficiency deteriorates due to mismatched processing power
Solution Approach 1:
The patent introduces processing power as a new parameter for storage device classification. Storage devices are divided into different sets based on their processing power levels, and data redistribution decisions are made by comparing processing power parameters between source and destination devices. This parameter-based approach allows efficient matching of data placement with device capabilities across heterogeneous storage pools.
Solution Approach 2:
The patent applies local quality by assigning different data types (hot data vs. cold data) to different storage device sets based on their local processing power characteristics. High-processing-power devices are designated for hot data requiring frequent access, while lower-processing-power devices handle cold data. This localized optimization ensures each storage device operates at its optimal performance level.
2Productivity
If data is redistributed across storage pools with different processing powers, then overall system performance is improved, but response time for individual operations increases due to heterogeneous device speeds
Solution Approach 1:
The patent uses processing power as a key parameter to predict and optimize response times. By evaluating the processing power of candidate storage devices before data redistribution, the system can select destination devices that will minimize response time for specific data types. This parameter-driven selection ensures that time-sensitive operations are routed to appropriately matched devices.
Solution Approach 2:
The patent optimizes response time locally by matching data access patterns with device processing capabilities. Hot data with high access frequency is placed on high-processing-power devices where response time is minimized, while cold data is placed on lower-processing-power devices. This local optimization strategy reduces overall response time loss by ensuring each data placement decision considers the specific performance characteristics of the target device.
3Adaptability or versatility
If storage devices are diversified into multiple pools, then adaptability of the storage system is improved, but load balancing deteriorates due to inability to span pools
Solution Approach 1:
The patent creates a universal data redistribution mechanism that works across diverse storage device types and pools. The processing power-based classification system serves as a common framework that can accommodate different device types (HDD, SSD, NVL) while maintaining consistent load balancing principles. This universal approach enables the system to adapt to various storage configurations while achieving effective load balancing across all pools.
Solution Approach 2:
The patent uses processing power as a standardized parameter that enables comparison and load balancing across heterogeneous storage devices. By converting diverse device types into a common processing power metric, the system can perform load balancing operations that span multiple storage pools. This parameter standardization allows the redistribution module to make informed decisions about data placement that optimize load distribution across the entire heterogeneous storage system.
Data Source
AI summary
Techniques involve: determining processing power of a plurality of storage devices in a plurality of storage pools, the storage devices in each of the storage pools having a same device type; dividing the plurality of storage devices into a plurality of sets based on the processing power, a difference in the processing power between the storage devices in each of the sets being below a predetermined threshold; and redistributing, among the plurality of sets, data stored in the plurality of storage devices based on workloads of the storage devices in the sets. Accordingly, load balancing of storage devices in the plurality of storage pools can be implemented, response speed of the storage system can be enhanced and storage resources in the plurality of storage pools can be exploited more sufficiently.