Modular Server Chassis Power Allocation for State-Based Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated power allocation techniques for compute servers in a chassis of a modular server require manual intervention to recover from system errors, such as a compute server becoming inactive due to a lack of power allocation response or insufficient power, leading to halted execution during boot processes.
Innovation Solution
A state-based automated power allocation system maintains a table of pending power requests and automatically grants power to compute servers when they transition from an inactive to a responsive state, while also reducing processor frequency when power thresholds are exceeded.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual intervention is used to recover from system errors in power allocation, then system reliability can be maintained through human oversight, but operational efficiency deteriorates due to halted execution and manual recovery steps
Solution Approach 1:
The system implements self-service through automated detection of compute server states and automatic granting of pending power requests. When a compute server transitions from an inactive state to a responsive state, the system automatically processes its pending power request without requiring manual intervention, thereby maintaining reliability while improving operational efficiency
Solution Approach 2:
The system performs preliminary action by maintaining a table of pending power requests that are prepared in advance. When a compute server becomes responsive again, the corresponding power request is already queued and can be immediately granted, eliminating the need for manual recovery steps and halting execution
2Productivity
If automated power allocation is implemented, then operational efficiency improves through automatic granting of power requests, but system complexity increases due to state monitoring and pending request management
Solution Approach 1:
The system manages complexity by performing preliminary action through the maintenance of a pending requests table. Power requests are captured and queued in advance, so when compute servers become responsive, the allocation can proceed automatically without complex real-time decision-making logic
Solution Approach 2:
The system uses feedback mechanisms by monitoring compute server states and using this information to automatically trigger power allocation. The state monitoring provides feedback that drives the automated granting process, creating a closed-loop system that improves efficiency while managing complexity through structured state-based control
3Speed
If power is allocated immediately upon request, then responsiveness improves, but power management efficiency deteriorates when compute servers are inactive or in error states
Solution Approach 1:
The system applies preliminary action by capturing and queuing power requests in a pending requests table before they are granted. This allows the system to prepare for rapid allocation when needed while avoiding wasteful power distribution to inactive or erroneous compute servers
Solution Approach 2:
The system uses feedback from compute server state monitoring to control power allocation timing. Power is allocated quickly when servers are responsive (maintaining speed) but withheld when servers are inactive or in error states (improving efficiency), creating an adaptive response that balances both requirements
Data Source
AI summary
Techniques are provided for state-based automated allocation of power to compute servers in a chassis of a modular server. One method comprises obtaining, for a modular server comprising a chassis with multiple compute servers, information characterizing pending requests for an allocation of power for the compute servers; monitoring a state of the compute servers; in response to a given compute server having a first designated state (e.g., the given compute server being non-responsive): obtaining at least a portion of the information characterizing a pending request for the allocation of power for the given compute server; determining whether the pending request for the allocation of power for the given compute server has been granted; and applying the requested allocation of power to the given compute server in response to the pending request being granted and the given compute server having a second designated state (e.g., the given compute server being responsive).