Dynamic Power Allocation via Measured Consumption Profiles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional blade server implementations are overly conservative in power management due to static power level allocations based on maximum power draw, leading to denial of power requests or performance throttling, despite sufficient power availability.
Innovation Solution
An information handling system with power profiling functionality that dynamically manages and allocates power based on actual consumption, using a power profiling module to transition resources to different power states and a management resource to allocate power accordingly, allowing for efficient allocation of scarce power resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If static power levels based on maximum power draw are used for power allocation, then power capacity is ensured not to be exceeded, but power allocation becomes overly conservative resulting in denial of power requests or forced performance throttling
Solution Approach 1:
The patent applies dynamics by transitioning from static power allocation based on maximum theoretical power draw to dynamic power allocation based on actual measured power consumption. The system continuously monitors real power usage and adjusts power allocation accordingly, allowing the power management to adapt to actual system conditions rather than relying on conservative fixed limits.
Solution Approach 2:
The patent implements feedback mechanisms by measuring actual power consumption and using this information to adjust power allocation decisions. The system receives feedback about real power usage from monitoring components and uses this feedback to dynamically adjust power allocation, preventing both power overruns and unnecessary throttling.
2Reliability
If static power levels based on maximum power draw are used, then power overconsumption is prevented, but sufficient power availability is not utilized leading to unnecessary throttling
Solution Approach 1:
The patent applies preliminary action by pre-characterizing power consumption for different operational states of information handling resources. By measuring and storing the power consumption associated with specific operational states beforehand, the system can quickly determine appropriate power allocation without needing to perform real-time measurements, enabling faster and more accurate power management decisions.
Solution Approach 2:
The patent changes the parameter basis for power allocation from fixed maximum power draw values to variable measured power consumption values. By using actual measured parameters that reflect real system behavior rather than theoretical maximums, the system optimizes power utilization while maintaining reliable control.
3Productivity
If measured power consumption values are used for dynamic power allocation, then power allocation efficiency is improved, but system complexity increases due to power profiling requirements
Solution Approach 1:
The patent reduces complexity by performing power consumption measurements and characterizations in advance during system setup or initialization. By pre-measuring and storing power consumption data for different operational states, the system avoids the need for complex real-time measurements during operation, simplifying the ongoing power management process while maintaining allocation efficiency.
Solution Approach 2:
The patent uses copying by creating a simplified power profile model that represents the power consumption characteristics of information handling resources. Instead of performing complex real-time power measurements, the system uses pre-created power consumption profiles that can be quickly referenced and applied to make power allocation decisions.
Data Source
AI summary
A method, system, and software instructions for allocating power in a information handling system are operable to respond to a power profiling request by transitioning a processing resource to a first power consumption state and obtaining and storing a first power consumption value. The first power consumption value is then retrieved and used to allocate power to the first processing resource in response to a power on request. The first power consumption state may be a state in which power consumption approximates a maximum power consumption. The processing resource may be further transitioned to a second power consumption state and a second power consumption value obtained. The second power consumption state may be a reduced performance state. Thereafter, responsive to determining that the system lacks sufficient power budget to fulfill a pending request for power, the processing resource is throttled and power is allocated using the second power consumption value.