Virtual Machine Power Scheduling With Reliability Index Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Monitoring and managing large numbers of virtual machines in cloud computing environments is inefficient, leading to wastefulness of resources and difficulty in maintaining specified service levels.
Innovation Solution
A system that monitors virtual machines for performance data, classifies them as active or idle based on idleness criteria, predicts periodic idle times, and transitions them between on and off states based on an on-off schedule to optimize resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If virtual machines are kept running to maintain service levels, then reliability is improved, but energy consumption increases
Solution Approach 1:
The system performs preliminary actions by predicting future idle periods of virtual machines based on historical performance data and usage patterns. Before the idle periods actually occur, the system schedules power-off actions during these predicted idle times, and schedules power-on actions before predicted busy periods. This preliminary scheduling allows the system to reduce energy consumption by turning off VMs during predicted idle times while maintaining service levels by ensuring VMs are running when actually needed.
2Use of energy by moving object
If virtual machines are turned off to conserve resources, then energy consumption is reduced, but reliability deteriorates
Solution Approach 1:
The system implements feedback mechanisms by continuously monitoring actual virtual machine usage and comparing it against predicted patterns. When actual usage deviates from predictions (indicating the VM was needed during a predicted idle period), the system learns from this feedback and adjusts future predictions and scheduling decisions. This feedback loop enables the system to improve prediction accuracy over time, thereby reducing false power-off events that would compromise service levels while maintaining energy savings during genuinely idle periods.
3Reliability
If manual monitoring and management of virtual machines is performed, then service levels can be maintained, but device complexity increases
Solution Approach 1:
The system enables self-service by automatically performing monitoring, analysis, and power scheduling operations without requiring manual intervention. The system autonomously collects performance data from virtual machines, analyzes usage patterns to predict idle periods, generates power schedules, and executes power-on/power-off operations. This automation eliminates the need for complex manual monitoring and management processes, reducing operational complexity while maintaining or improving service levels through consistent, data-driven decision-making.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer monitors virtual machines that execute on physical machines, and extracts performance data. It classifies the virtual machines as active or idle during an observation time. Based on the performance data and on idleness criteria, the computer generates an active-idle series for each virtual machine, and determines at least one periodicity of recurring idle times. Then, for each virtual machine with the at least one periodicity, the computer determines an on-off schedule, and transitions the virtual machines between an on state and an off state in accordance with the schedule. The computer also performs a frequency analysis of the active-idle series and compares an active count percentage to a frequency threshold. The computer further calculates a reliability index of the schedule during a reliability estimation time that belongs to the observation time. In case that the reliability index goes below a threshold, the computer repeats the frequency analysis by applying an adjusted frequency threshold.