Chassis Power Balancing for Blade Servers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face challenges in efficiently allocating and balancing power across blade servers, leading to performance degradation due to excessive power allocation and inefficient power usage, as traditional methods like power capping restrict peak performance.
Innovation Solution
Implementing a chassis management controller that redistributes power based on blade server priority and actual consumption, using policy information to allocate a power budget and dynamically adjust power distribution among servers, ensuring fair and efficient power usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If power capping is used to reduce power consumption, then electricity costs are reduced, but server peak performance is degraded
Solution Approach 1:
The patent implements dynamic power allocation where the chassis management controller continuously monitors power consumption of individual blade servers and adjusts power distribution in real-time. Instead of static power capping, the system dynamically reallocates power from underutilized servers to those requiring peak performance, resolving the contradiction between reducing overall power consumption and maintaining individual server performance capabilities.
Solution Approach 2:
The system changes the power allocation parameters for each blade server based on actual usage patterns and performance requirements. By monitoring power consumption metrics and performance indicators, the chassis management controller adjusts power distribution parameters dynamically, allowing servers to operate at peak performance when needed while reducing total power consumption during low-utilization periods.
2Productivity
If more power is allocated to blade servers, then peak performance is maintained, but unnecessary power wastage increases
Solution Approach 1:
The chassis management controller implements a self-service power allocation system that automatically monitors each blade server's power consumption and performance needs, then autonomously reallocates power without manual intervention. The system identifies servers consuming excessive power relative to their utilization and redirects that power to servers requiring peak performance, eliminating power wastage while maintaining productivity.
Solution Approach 2:
The patent employs continuous feedback loops where the chassis management controller monitors power consumption metrics and performance indicators from each blade server. Based on this feedback, the system adjusts power allocation in real-time, reducing power to underutilized servers and increasing power to those requiring peak performance, thereby eliminating wastage while maintaining necessary productivity levels.
3Device complexity
If static power allocation is used, then power distribution is simple, but it cannot adapt to changing server workload demands
Solution Approach 1:
The patent transforms static power allocation into a dynamic system where the chassis management controller continuously monitors workload demands and power consumption across all blade servers. The system automatically adjusts power distribution in real-time based on changing conditions, providing adaptability to varying server demands while managing complexity through automated control algorithms.
Solution Approach 2:
The chassis management controller serves multiple functions: it monitors power consumption, tracks performance metrics, identifies allocation inefficiencies, and executes power reallocation decisions. This multi-functional approach handles both the simplicity of centralized control and the complexity of adaptive power management within a single integrated system.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Techniques are provided to redistribute and rebalance power to a plurality of blade servers with a chassis unit. At a chassis management controller device in a chassis unit comprising a plurality of blade server devices, policy information is stored for the chassis unit. The policy information comprises power limit information that indicates a power budget for the chassis unit and blade server priority information that indicates priority for access to power for each of the plurality of blade servers in the chassis unit. Power is distributed to the plurality of blade servers based on the blade server priority information for each of the plurality of blade servers and the power limit information for the chassis unit. The actual power consumption of each of the plurality of blade servers is determined and power is redistributed to the plurality of blade servers based on the actual power consumption and the policy information.