Service Redeployment Across Power Modes Under Latency Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computing systems face challenges in efficiently redeploying services based on power consumption needs while ensuring latency requirements are met, leading to potential overconsumption of energy.
Innovation Solution
A system and method for redeploying services based on power consumption needs by analyzing a call graph to identify services that can be moved to lower power mode systems, considering statelessness and latency requirements, and utilizing a service power manager to optimize energy efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If services are redeployed to computing systems operating at lower power modes, then energy consumption is reduced, but latency requirements may not be satisfied
Solution Approach 1:
The patent applies local quality by assigning different power modes to different computing systems based on their specific roles and service requirements. Each computing system operates at an optimized power mode (first, second, or third power mode) tailored to its function, allowing the network to achieve overall energy efficiency while maintaining latency requirements for time-sensitive services.
Solution Approach 2:
The patent changes the power mode parameter of computing systems dynamically. The service power manager can transition computing systems between different power modes (first power mode with highest performance, second power mode with reduced power consumption, third power mode with lowest power consumption) based on service requirements and energy efficiency goals, thereby resolving the contradiction between energy consumption and latency satisfaction.
2Use of energy by moving object
If services are redeployed across multiple computing systems, then energy efficiency is optimized, but system complexity increases
Solution Approach 1:
The patent implements a universal service power manager that can manage service deployment across multiple computing systems with different power modes. This single management entity handles power mode selection, service deployment decisions, and energy efficiency optimization across the entire network, reducing the complexity burden on individual components while achieving system-wide energy optimization.
Solution Approach 2:
The service power manager acts as an intermediary between services and computing systems. It makes intelligent decisions about service deployment and power mode selection, shielding individual computing systems from the complexity of power mode management while enabling energy-efficient operation across the network.
3Reliability
If all services are kept on high-power computing systems, then latency requirements are satisfied, but energy consumption increases
Solution Approach 1:
The patent segments services into different categories based on their latency requirements and power consumption characteristics. Time-sensitive services are deployed on computing systems operating at first or second power modes, while energy-intensive services with relaxed latency requirements are deployed on systems at third power mode. This segmentation allows the network to satisfy latency requirements for critical services while minimizing overall energy consumption.
Solution Approach 2:
The patent applies partial action by deploying services selectively on computing systems operating at different power modes rather than all services on high-power systems. Only services requiring low latency are deployed on first power mode systems, while other services can tolerate deployment on lower power mode systems, thereby reducing total energy consumption while maintaining necessary performance levels.
Data Source
AI summary
Provided are a computer program product, system, and method for redeploying services in computing systems based on power consumption needs of the services. A call graph indicating multiple calls between interlinked services. A determination is made of a first service in the call graph consuming energy at a first service power mode in one of the computing systems. A determination is made of a second service, in one of the computing systems, operating at a second service power mode requiring less energy than a power mode of the computing system including the second service. The second service is in a subgraph of the call graph extending from the first service. An alternative computing system is determined, comprising one of the computing systems not including the second service, operating at the second service power mode. The second service is redeployed to the alternative computing system.