Storage Virtualization for Power Minimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face significant energy consumption challenges due to exponential data growth, with storage being a major consumer of power, and existing power management solutions often require expensive data migration and are not transparent to applications.
Innovation Solution
The implementation of dynamic storage virtualization techniques that map workload to minimize power consumption by creating multiple physical storage volumes with varying performance and power levels, predicting I/O access behavior, and remapping virtual to physical volumes without data migration, allowing for transparent power management across heterogeneous storage devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by stationary object
If data migration is used to manage storage power consumption, then power consumption can be reduced, but data migration costs increase and application transparency is lost
Solution Approach 1:
The patent introduces a storage virtualization layer as an intermediary between applications and physical storage devices. This virtualization layer abstracts the complexity of data migration from applications, allowing power management operations to be performed on virtual storage volumes rather than requiring direct data migration. The virtualization layer handles the mapping and remapping of virtual volumes to physical volumes, eliminating the need for application-level data migration while still achieving power consumption reduction.
Solution Approach 2:
The patent segments storage management into multiple independent virtual storage volumes that can be individually mapped to different physical storage devices. This segmentation allows selective remapping of specific virtual volumes to different physical volumes based on power consumption characteristics, rather than requiring migration of all data. Each virtual volume can be independently managed and remapped without affecting other volumes, reducing overall migration complexity.
2Use of energy by stationary object
If multiple physical storage volumes are created with varying performance levels, then power consumption is reduced, but storage system complexity increases
Solution Approach 1:
The patent creates multiple physical storage volumes with different performance and power characteristics, but manages them through a universal virtualization layer that presents a unified interface to applications. The virtualization layer handles the complexity of managing heterogeneous physical volumes by providing consistent virtual volume interfaces. This allows the system to utilize diverse storage resources (different performance levels, power consumption profiles) while maintaining simple, uniform access methods for applications.
Solution Approach 2:
The patent changes the mapping parameters between virtual and physical volumes dynamically based on workload characteristics and power consumption requirements. By adjusting the virtual-to-physical volume mapping configuration rather than changing the physical storage hardware itself, the system can adapt to different power management scenarios. The remapping process changes the association parameters between virtual and physical volumes, allowing flexible power management without permanent hardware changes.
3Use of energy by stationary object
If storage volumes are dynamically reconfigured based on workload, then power consumption decreases, but data consistency challenges arise
Solution Approach 1:
The patent performs preliminary actions by creating multiple physical storage volumes in advance with different performance and power characteristics before workload changes occur. The virtualization layer pre-configures the mapping relationships between virtual and physical volumes, so when remapping is needed for power management, the target physical volumes are already prepared and available. This preliminary preparation reduces the risk of data consistency issues during dynamic reconfiguration, as the system doesn't need to create or initialize volumes on-the-fly during workload changes.
Data Source
AI summary
Techniques for minimizing storage power consumption are provided. The techniques include generating one or more physical storage volumes and one virtual storage volume for each physical storage volume, creating a mapping from virtual storage volumes to physical storage volumes, determining input/output (I/O) access behavior of one or more applications using statistical analysis, and re-mapping the virtual to physical volume mapping based on the determined I/O access behavior of the one or more applications to minimize storage power consumption while meeting a required performance.