Hardware Resource Capacity Reporting for Power-Efficient Workload Placement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing datacenter management systems face inefficiencies in power usage, as they either consolidate workloads to save power or enforce power caps, lacking effective methods to optimize power usage across hardware resources.
Innovation Solution
A method and system that utilizes a baseboard management controller to adjust available resource capacity based on power efficiency, ensuring that workload placement decisions are made considering only resources that maintain a setpoint power efficiency, thereby optimizing power usage without altering existing workload placement logic.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If workload consolidation is implemented to save power, then power consumption is reduced, but system reliability and workload distribution deteriorate
Solution Approach 1:
The patent changes the parameter representation of available capacity from absolute values to efficiency-adjusted values. The baseboard management controller reports available capacity only when it can be provided without dropping below a threshold power efficiency, effectively filtering capacity reports based on power efficiency parameters while maintaining system reliability
Solution Approach 2:
The baseboard management controller acts as an intermediary between the hardware resources and the workload placement system. It adjusts the reported available capacity based on power efficiency calculations, providing filtered information that enables power-efficient decisions without compromising system reliability
2Use of energy by stationary object
If power caps are enforced on servers, then power budget is controlled, but resource utilization and productivity deteriorate
Solution Approach 1:
Instead of static power caps, the system dynamically adjusts the reported available capacity based on current power efficiency conditions. The baseboard management controller continuously monitors and reports capacity availability that maintains power efficiency above thresholds, enabling flexible resource allocation without rigid power constraints
Solution Approach 2:
The system performs preliminary power efficiency evaluation before workload placement decisions are made. The baseboard management controller pre-calculates and reports only the capacity that can be provided while maintaining power efficiency thresholds, preventing inefficient placements before they occur
3Loss of energy
If existing workload placement logic is altered to optimize power efficiency, then power efficiency improves, but system complexity and ease of operation deteriorate
Solution Approach 1:
The patent extracts the power efficiency calculation and adjustment logic from the main workload placement system and places it in the baseboard management controller. This separation allows the workload placement logic to remain simple while power efficiency optimization is handled by the intermediary controller that adjusts capacity reports
Solution Approach 2:
The baseboard management controller autonomously adjusts the reported available capacity based on power efficiency calculations without requiring changes to the workload placement system. The controller serves itself by filtering and reporting only the capacity that meets power efficiency requirements, eliminating the need for complex integration
Data Source
AI summary
A method includes identifying hardware resource requirements of a candidate workload instance, wherein the hardware resource requirements of the workload instance include a required capacity for each of a plurality of hardware resource types. The method further includes identifying, for each of the plurality of hardware resource types required by the candidate workload instance, a capacity of a hardware resource of the hardware resource type that is currently available on a server without causing a power efficiency of the hardware resource to decline below a setpoint power efficiency; and determining, for each of the plurality of hardware resource types required by the candidate workload instance, whether the identified capacity of the hardware resource on the server is greater than or equal to the required capacity for the hardware resource type. The determination may support a decision to place the workload instance on the server or another server.