Demand Based Power Re-allocation for Server Throttling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In server systems, there is no mechanism for power re-balancing post-allocation, leading to some servers being continuously throttled and running at lower performance while others have surplus power due to uneven power distribution.

Innovation Solution

A demand-based power re-allocation system that assigns power levels by priority, throttles underperforming servers, tracks power usage, and reallocates excess power from unthrottled servers to those in need, using a chassis management controller and remote access controller to manage and communicate power adjustments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If power is allocated to servers based on initial distribution, then some servers receive sufficient power, but other servers are continuously throttled and cannot achieve maximum performance

Engineering Contradiction:
Improveserver performanceVSAvoidpower throttling
Core Design Contradiction:
ProductivityVSLoss of energy

Solution Approach 1:

The system dynamically adjusts power allocation among servers based on real-time power consumption monitoring and demand assessment. The chassis management controller continuously monitors power usage and reallocates power from underutilized servers to those requiring additional power, enabling adaptive power distribution that responds to changing server demands and eliminates continuous throttling of any single server.

Inventive Principle:
Principle #15Dynamics

2Productivity

If power is allocated to ensure some servers run at maximum performance, then those servers achieve high productivity, but other servers suffer from power throttling and cannot operate optimally

Engineering Contradiction:
Improveoverall system productivityVSAvoidpower management
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The system implements self-service power management where the chassis management controller automatically monitors power consumption across all servers, identifies underutilized servers with excess power capacity, and reallocates power to servers experiencing throttling. This automated self-balancing mechanism eliminates the need for manual power management intervention while optimizing overall system productivity and ensuring fair power distribution.

Inventive Principle:
Principle #25Self-service

3Adaptability or versatility

If power allocation is fixed after initial distribution, then power management is simple, but servers cannot adapt to changing power demands and some remain continuously throttled

Engineering Contradiction:
Improvepower adaptabilityVSAvoidpower management system
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system employs feedback mechanisms where the chassis management controller continuously monitors power consumption levels of individual servers, compares actual usage against allocated power, and automatically triggers power reallocation when imbalances are detected. This closed-loop feedback system enables dynamic adaptation to changing power demands while maintaining manageable complexity through automated decision-making algorithms that respond to real-time system state.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS8713334B2Demand based power allocation
Publication Date: 2014.04.29 DELL PROD LP
  • US8713334B2 patent drawing
  • US8713334B2 patent drawing
  • US8713334B2 patent drawing

AI summary

A demand based power re-allocation system includes one or more subsystems to assign a power allocation level to a plurality of servers, wherein the power allocation level is assigned by priority of the server. The system may throttle power for one or more of the plurality of servers approaching the power allocation level, wherein throttling includes limiting performance of a processor, track server power throttling for the plurality of servers. The method compares power throttling for a first server with power throttling for remaining servers in the plurality of servers and adjusts throttling of the plurality of servers, wherein throttled servers receive excess power from unthrottled servers.