Power Capping in Composable Computing Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In composable computing systems, there is no existing provision to communicate power capping or throttling events to remotely-shared resources, leading to potential power wastage as these resources continue to run at full capacity even when demand is scaled down.
Innovation Solution
A computer program product that monitors performance metrics within a composable computing system, identifies service level agreements, and imposes power caps on compute and associated hardware resources by sending commands to baseboard management controllers, allowing for adaptive power capping to ensure efficient power usage while maintaining performance standards.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If power capping is applied to compute resources in composable computing systems, then power consumption is reduced, but remotely-shared hardware resources cannot receive power capping commands and continue to run at full power causing power wastage
Solution Approach 1:
The patent introduces a power capping intermediary layer that sits between the compute resource power management system and the remotely-shared hardware resources. This intermediary receives power capping events from the compute resource and translates them into appropriate power capping commands for the hardware resources, enabling coordinated power management across distributed power domains without requiring direct integration between all components.
2Adaptability or versatility
If hardware resources are disaggregated into separate power domains, then resource flexibility and composable architecture are improved, but communication of power capping events between domains becomes impossible
Solution Approach 1:
The patent implements a universal power capping communication mechanism that can operate across different power domains and hardware resource types. The system uses a standardized interface and message format that allows power capping events to be transmitted from compute resources to various types of hardware resources (storage, memory, graphics) even when they are physically located in separate power domains with independent power control.
Data Source
AI summary
A computer program product provides program instructions that are executable by a processor to cause the processor to perform various operations. The operations may include monitoring a performance metric for a workload instance being executed by a composed system within a pool of composable resources in a composable computing system. The composed system includes a compute resource and an associated hardware resource selected from a data storage resource, a memory resource and/or a graphic processing resource. A service level agreement is identified for the workload instance, wherein the agreement includes a minimum level of the performance metric that the composed system must provide to support the workload instance. A power cap may be imposed on the compute resource, and a power cap may be imposed on the associated hardware resource by sending a power capping command to a baseboard management controller on a server including the associated hardware resource.