Virtual Machine Idle Detection via Cluster Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques fail to accurately and consistently identify idle virtual machines in data centers, leading to unnecessary resource wastage and increased costs due to the inability to determine which virtual machines are no longer serving a purpose.
Innovation Solution
A computing device collects utilization metrics such as processing, disk, and network usage over time, separates them into training and validation sets, and uses cluster analysis to determine whether a virtual machine is idle, allowing for the identification and shutdown of idle machines to conserve resources and reduce costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional techniques are used to manage virtual machines, then ease of operation is maintained, but measurement precision of idle virtual machine detection deteriorates
Solution Approach 1:
The patent introduces cluster analysis as an intermediary technique that processes utilization metrics and translates them into idle virtual machine identification. This mediator layer converts complex multi-dimensional metric data into actionable insights without requiring direct complex detection logic, thereby improving measurement precision while managing system complexity through a specialized analytical component.
Solution Approach 2:
The patent replaces conventional manual or rule-based detection mechanisms with data-driven cluster analysis. By substituting mechanical detection approaches with statistical pattern recognition on utilization metrics, the system achieves higher measurement precision in identifying idle virtual machines without relying on complex manual intervention or simplistic threshold-based rules.
2Measurement precision
If cluster analysis is applied to utilization metrics, then measurement precision of idle detection improves, but device complexity increases
Solution Approach 1:
The patent segments the detection process into distinct phases: data collection from multiple sources, preprocessing of utilization metrics, cluster analysis execution, and result interpretation. This segmentation breaks down the complex analytical task into manageable components, improving measurement precision through systematic processing while controlling device complexity by distributing computational workload across separate functional modules.
Solution Approach 2:
The patent performs preliminary actions by collecting and preprocessing utilization metrics before applying cluster analysis. By preparing the data in advance—gathering processing, memory, storage, and network metrics and organizing them for analysis—the system improves the precision of idle detection while reducing the computational complexity during the actual cluster analysis phase, as the heavy lifting of data preparation is completed beforehand.
3Loss of energy
If idle virtual machines are not identified, then ease of operation is maintained, but loss of energy increases due to unnecessary resource consumption
Solution Approach 1:
The patent employs cluster analysis as an intermediary that bridges the gap between raw utilization metrics and idle virtual machine identification. This mediator processes multiple types of computing resource metrics (CPU, memory, storage, network) and translates them into reliable idle status determination, thereby reducing energy loss from wasted resources while managing the difficulty of detection through systematic analytical processing rather than ad hoc methods.
Solution Approach 2:
The patent implements a universal detection approach that handles multiple types of virtual machines and multiple resource metrics through a single cluster analysis framework. This multi-functional system can identify idle virtual machines across different workloads and resource types, reducing energy waste comprehensively while avoiding the need for separate detection mechanisms for each machine type, thereby addressing the detection difficulty through a unified solution.
Data Source
AI summary
The detection of utilized virtual machines through usage pattern analysis is described. In one example, a computing device can collect utilization metrics from a virtual machine over time. The utilization metrics can be related to one or more processing usage, disk usage, network usage, and memory usage metrics, among others. The utilization metrics can be used to determine a number of clusters, and the clusters can be used to organize the utilization metrics into groups. Depending upon the number or overall percentage of the utilization metrics assigned to individual ones of the plurality of clusters, it is possible to determine whether or not the virtual machine is a utilized or an idle virtual machine. Once identified, utilized virtual machines can be migrated in some cases. Idle virtual machines can be shut down to conserve processing resources and costs in some cases.