Blade Server Power Management with Scalable PSU Configuration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face significant energy wastage due to idle servers, as current power management techniques are not practical for rapid transitions and brief intervals of activity, and existing solutions like CPU throttling and energy-proportional computing are limited in effectiveness for server systems.
Innovation Solution
A power management method that transitions networked computers from an active state to a low power state when idle, maintaining the network interface and timer active, and uses a load-sharing power supply system with multiple power supply units that can be selectively connected or isolated to efficiently manage power usage based on demand.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If servers remain powered on during idle periods, then they can rapidly resume operations, but energy is wasted continuously
Solution Approach 1:
The system dynamically transitions servers between active and low-power states based on workload conditions. The power management method monitors server activity and automatically switches power states, making the power consumption adaptive rather than static. This resolves the contradiction by allowing servers to be dynamic in their power state rather than fixed in one state.
Solution Approach 2:
The invention changes the power state parameter of servers based on operational needs. By transitioning from full power to low-power state and back, the system modifies the power consumption parameter dynamically. This parameter change enables energy savings during idle periods while maintaining the capability to quickly return to full operation when needed.
2Loss of energy
If servers are consolidated to reduce idle systems, then total server count decreases, but peak demand capacity requirements remain
Solution Approach 1:
The system provides dynamic power management that allows servers to switch between active and low-power states based on real-time workload demands. This dynamic capability enables the system to handle peak demands by activating additional servers while saving energy during low-utilization periods, thus maintaining adaptability without continuous energy consumption.
Solution Approach 2:
The power management system provides multi-functionality by serving both energy conservation and peak demand capacity requirements through the same infrastructure. The ability to rapidly transition between power states makes the server infrastructure universal in handling both idle and peak conditions without requiring separate systems.
3Speed
If power states are transitioned rapidly, then responsiveness is improved, but power management complexity increases
Solution Approach 1:
The power management system operates autonomously by monitoring server activity and automatically transitioning power states without manual intervention. The system serves itself by detecting idle conditions and initiating power state changes, which simplifies operation despite the rapid transitions. This self-service approach handles the complexity internally while presenting a simple interface.
Solution Approach 2:
The system uses feedback from workload monitoring to control power state transitions. By continuously monitoring server activity and using this feedback to determine when to transition power states, the system achieves rapid responsiveness while managing complexity through closed-loop control. The feedback mechanism ensures transitions occur at appropriate times based on actual conditions.
Data Source
AI summary
A power management method for use by a blade server or other networked computer having a CPU, memory, network interface, and timer. The method includes monitoring processes in the computer, determining that the computer has become idle, and switching the CPU and memory from an active state to a low power state while maintaining the network interface and timer in an active state. Power management for multicore processors is also provided. A power supply system can be used for providing shared power to groups of the networked computers that each have a similar active power demand. Each PSU in the power supply system is capable of supplying the active power demand at an efficiency greater than a preselected desired minimum efficiency. Changes in power requirements resulting from a computer switching between low power and active states can be accommodated by switching a single PSU into or out of the system.