Data Center Load Assignment Using Air Conditioning Sensitivity

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for managing task loads in data centers fail to achieve comprehensive power saving, as they often result in local power savings rather than overall reductions, and require complex calculations or separate servers for evaluating power saving indices.

Innovation Solution

A method that assigns task loads to servers based on air conditioning sensitivities, prioritizing servers with the lowest maximum change rates in temperature, thereby optimizing cooling power usage and reducing the intake temperature of air to servers, which minimizes the cooling power required by the air conditioner.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Use of energy by stationary object

If task loads are assigned to servers without considering air conditioning sensitivity, then load distribution is simple, but cooling power consumption increases

Engineering Contradiction:
Improvecooling power consumptionVSAvoidload assignment complexity
Core Design Contradiction:
Use of energy by stationary objectVSDevice complexity

Solution Approach 1:

The patent pre-calculates air conditioning sensitivity coefficients for each server before load assignment. These coefficients represent how much each server's heat generation affects the cooling load. By having these coefficients ready in advance, the system can quickly determine optimal load assignments without complex real-time calculations, thus reducing cooling power consumption while maintaining simple operational complexity.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent transforms the load assignment problem from a complex optimization problem into a simpler parameter-based selection problem. By using air conditioning sensitivity coefficients as key parameters, the system can directly select servers for load assignment based on these pre-determined values, avoiding complex iterative calculations and achieving energy savings with minimal computational overhead.

Inventive Principle:
Principle #35Parameter changes

2Use of energy by stationary object

If comprehensive power saving is achieved through complex calculations, then power consumption reduces, but calculation complexity and time increase

Engineering Contradiction:
Improveoverall power consumptionVSAvoidcalculation time
Core Design Contradiction:
Use of energy by stationary objectVSLoss of time

Solution Approach 1:

The patent performs air conditioning sensitivity calculations in advance, before actual load assignment is needed. This preliminary calculation captures the thermal characteristics of each server and its impact on cooling requirements. When load assignment is needed, the system simply uses these pre-computed coefficients rather than performing complex thermal simulations in real-time, thus achieving comprehensive power saving without significant calculation time loss.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If task loads are assigned to minimize cooling power, then cooling efficiency improves, but the system requires complex evaluation of power saving indices

Engineering Contradiction:
Improvecooling efficiencyVSAvoidpower saving evaluation complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent simplifies the power saving evaluation by transforming it into a parameter-based selection process using air conditioning sensitivity coefficients. Instead of evaluating complex power saving indices that require multiple variables and iterative calculations, the system uses pre-determined sensitivity coefficients to directly identify optimal servers for load assignment, thereby improving cooling efficiency while reducing evaluation complexity.

Inventive Principle:
Principle #35Parameter changes

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach allows for easy assignment of task loads, reducing the overall power consumption in the data center by optimizing the assignment of loads, thereby achieving comprehensive power saving without the need for complex calculations or additional servers.

Implementation Method 1

an air conditioner for cooling, which cools air taken in by a plurality of servers

Methodology Applied
Scientific EffectHeat exchange: Heat Exchanger

Implementation Method 2

a plurality of servers cooled by heat radiation

Methodology Applied
Scientific EffectHeat radiation: Thermal Radiation

Data Source

PatentEP2575003B1Method for determining assignment of loads of data center and information processing system
Publication Date: 2014.05.14 HITACHI LTD
  • EP2575003B1 patent drawingFigure 1A~1C
  • EP2575003B1 patent drawingFigure 2
  • EP2575003B1 patent drawingFigure 3

AI summary

A load management system that is provided for a data center and determines assignment of task loads so as to achieve power saving in a whole computer room easily assigns task loads to information processing devices. A method for determining assignment of loads of a data center, the data center including a plurality of servers cooled by heat radiation, in a room isolated from an outdoor space, that allows air to be taken into and discharged from the room, wherein for the assignment of a plurality of processes to each server, when proportionality coefficient (Aij) indicates the ratio of temperature of air taken in the servers (j) arranged in the room to a load on the server (i) arranged in the room, and when the server (i) is compared with the respective servers (j) for the proportionality coefficient (Aij) to obtain the maximum proportionality coefficients (Ai-max), the plurality of processes are assigned to the plurality of servers in order from a process applied with the highest load but in order from a server with a proportionality coefficient that is smallest among the maximum proportionality coefficients (Ai-max).