Dynamic Service Distribution via Resource Capacity Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for dynamically distributing services in computer networks often fail to respond effectively to changing resource demands, leading to occasional resource bottlenecks and uncontrollable declines in service performance, despite predictive approaches that cannot guarantee meeting service requirements.
Innovation Solution
A method that derives the past resource capacity required for each service, predicts future resource capacity using statistical evaluation and optimization criteria, and redistributes services to ensure sufficient resources are available while minimizing the number of computers used, allowing for energy-saving modes and secure distribution guarantees.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If services are redistributed based on currently observed demand pattern, then resource distribution adapts to current conditions, but the system cannot respond in time to changing demand patterns leading to resource bottlenecks
Solution Approach 1:
The system performs preliminary actions by predicting future resource demand patterns before they actually occur. The prediction mechanism analyzes historical demand data and proactively determines future resource capacity requirements, allowing the system to redistribute services in advance to prevent resource bottlenecks before they happen, thus resolving the time delay issue while maintaining adaptability.
2Reliability
If more computers are used to ensure sufficient resources, then service requirements are met, but energy consumption increases
Solution Approach 1:
The system dynamically adjusts the number of active computers based on predicted resource demand. Instead of maintaining a static configuration that always ensures sufficient resources (consuming more energy), the system continuously updates its prediction of required resource capacity and activates or deactivates computers accordingly, thus maintaining service reliability while minimizing energy consumption by keeping only the necessary number of computers active.
3Loss of time
If predictive approaches are used to forecast future resource demand, then forward planning is enabled, but service requirements cannot be guaranteed to be met
Solution Approach 1:
The system implements feedback by continuously monitoring actual resource demand and comparing it with predicted values. When deviations are detected, the system adjusts its prediction model and redistributes services to ensure service requirements are met. This closed-loop approach maintains forward planning capability while guaranteeing reliability by correcting prediction errors in real-time.
4Loss of energy
If services are consolidated on fewer computers, then energy consumption is reduced, but resource bottlenecks may occur
Solution Approach 1:
The system performs preliminary action by predicting future resource capacity requirements before consolidation is executed. It calculates the optimal number of computers needed based on predicted demand, ensuring that service requirements will still be met even with reduced resource capacity. This allows aggressive energy-saving consolidation while maintaining reliability through advance planning and verification.
Data Source
AI summary
The invention relates to a method for dynamically distributing one or more services in a network comprised of a plurality of computers. According to certain aspects of the invention, a past chronological progression of a resource capacity required for a respective service according to a prescribed service requirement is derived from a past chronological progression of the resource demand for the respective service in a predetermined time interval. The past chronological progression of the resource capacity required for the respective service is then used to predict a chronological progression of the resource capacity required for the respective service. The services performed on the computers are finally distributed based on one or more optimization criteria, including that the respective computers provide enough resources for the services performed on the respective computers based on the predicted chronological progressions of the resource capacities required for the respective services.