Hyper-Converged Storage Power Management via Priority-Based State Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In hyper-converged systems, moving virtual machines (VMs) between nodes to reduce power consumption degrades performance due to the need for data access between nodes, and storage devices may not have enough empty space to store all data, leading to inefficiencies in power management.
Innovation Solution
A processor unit determines the power state of components based on their priority and data characteristics to optimize power consumption within set limits, allowing for reduced power usage without degrading performance by adjusting the power states of processors and storage devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If VM is moved between nodes to reduce power consumption, then power consumption is reduced, but performance degrades due to inter-node data access
Solution Approach 1:
The patent applies local quality by differentiating between local storage (high performance, high power) and remote storage (lower performance, lower power). The system dynamically selects access modes based on data location: local access for frequently accessed data and remote access for less critical data, optimizing the trade-off between performance and power consumption at each node
Solution Approach 2:
The system dynamically adjusts power states and access modes based on real-time conditions. The power management unit monitors workload, data access patterns, and power consumption, then dynamically transitions components between active and low-power states, and switches between local and remote access modes to optimize the performance-power trade-off
2Adaptability or versatility
If data moves with VM between nodes, then storage capacity is utilized, but destination storage device may not have enough empty space
Solution Approach 1:
The patent introduces a power management unit and control system as intermediaries that coordinate data migration and power state transitions. The control unit monitors storage capacity across nodes and manages the migration process, selecting appropriate destination nodes with sufficient space and coordinating the transfer to ensure storage availability is maintained
Solution Approach 2:
The system implements a universal storage pool that can be dynamically allocated across multiple nodes. Storage resources are abstracted and managed as a unified pool, allowing flexible allocation and migration of data between nodes based on capacity availability and power management requirements
Data Source
AI summary
Computer components, such as processors and storage devices, provide a performance and consumes an electric power within a range of an upper limit performance and an upper limit power consumption of a power state set for the component among a plurality of power states corresponding to a type of the component. A processor unit determines whether a budget power as a power consumption permitted for a target computer is equal to or more than a power consumption of the target computer or not. When the determination result is false, for at least one component of the target computer, the processor unit selects a power state based on at least one of a priority of an operation using the component and a data characteristic corresponding to the component among a plurality of types of power states corresponding to a type of the component as power state of the component.