Dynamic Power Allocation for Data Center Server Racks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face inefficiencies in power management due to reserved capacity for redundancy, leading to unused or unsold power capacity, which is not effectively utilized or monetized.
Innovation Solution
A controller system that interrogates power supply and operational status systems to determine if power demands can be met, and if not, commands server racks to self-configure and operate at reduced or increased power supply levels based on tiered availability contracts, optimizing power usage and availability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If power capacity is reserved for redundancy, then reliability is improved, but power utilization efficiency deteriorates
Solution Approach 1:
The patent implements dynamic power allocation where the controller continuously monitors power availability and adjusts the power supply level to server racks in real-time. Instead of static reserved capacity, the system dynamically scales power distribution based on current conditions, allowing previously reserved power to be utilized when available while maintaining reliability through adaptive adjustment.
Solution Approach 2:
The system changes the power supply parameter from a fixed reserved level to a variable level based on service level agreements and real-time power availability. The controller adjusts power allocation parameters dynamically, transforming the static redundancy model into a flexible parameter-driven approach that optimizes both reliability and utilization.
2Productivity
If power capacity is increased to serve more customers, then customer serviceability is improved, but power supply cost increases
Solution Approach 1:
The patent implements partial power allocation where server racks receive power at varying levels based on their service level agreements rather than full capacity allocation. This allows the data center to serve more customers with partial power provision, and dynamically allocate additional power when available, thereby increasing customer serviceability without proportionally increasing total power supply capacity and cost.
3Loss of energy
If tiered power allocation is implemented, then power utilization efficiency is improved, but system complexity increases
Solution Approach 1:
The patent segments the power allocation system into discrete service level agreements with specific power levels, allowing differentiated power distribution to different server racks. This segmentation enables efficient power utilization by matching supply levels to demand priorities while maintaining manageable system complexity through standardized tiers rather than continuous variable control.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In some examples, a method includes acquiring status information of at least one resource of a data center, estimating, based on the status information, whether load on a power supply will exceed a threshold value at a time subsequent a time of the estimating, and in response to estimating that load on the power supply will exceed the threshold value at the time subsequent the time of the estimating, transmitting a command signal to a server rack that is coupled to the power supply to configure equipment of the server rack to operate at a reduced power supply level that is reduced from a power supply level at the time of the estimating and that is based on a service level agreement associated with the server rack.