Virtual Machine Utilization Detection via Distributed ML Classification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing virtual machine management systems struggle to efficiently identify and optimize underutilized virtual machines, leading to increased operating and capital expenditures due to resource wastage, especially in large-scale deployments where hundreds of thousands of virtual machines require accurate and timely utilization assessment.

Innovation Solution

A system and method that utilize a distributed data architecture to assess virtual machine operating characteristics, incorporating machine learning and pattern recognition techniques to classify virtual machine utilization, allowing for dynamic resource allocation, consolidation, and de-provisioning of idle machines, thereby optimizing resource usage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional virtual machine management systems are used to identify underutilized virtual machines, then comprehensive monitoring of all virtual machines can be achieved, but the computing power and time required for classification increases significantly

Engineering Contradiction:

Improveaccuracy of underutilized virtual machine identificationVSAvoidcomputing power and time required for classification

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the virtual machine management system into multiple specialized components: data collection agents that gather metrics from individual virtual machines, a data processing layer that aggregates and normalizes information, and a classification engine that applies machine learning models. This segmentation allows parallel processing of large numbers of virtual machines, reducing overall computation time while maintaining identification accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary actions by pre-processing and normalizing data from multiple sources before classification. It collects baseline performance metrics, establishes normalization rules, and prepares data structures in advance, so that when classification is needed, the computationally intensive work has already been partially completed, reducing the actual classification time.

Inventive Principle:

Principle #10Preliminary action

2Loss of energy

If traditional virtual machine management systems are used, then resource allocation can be monitored, but resource wastage due to underutilization increases operating and capital expenditures

Engineering Contradiction:

Improveresource wastage from underutilized virtual machinesVSAvoidoperating and capital expenditures

Core Design Contradiction:

Loss of energyVSProductivity

Solution Approach 1:

The system implements continuous feedback loops where performance metrics are collected from virtual machines, analyzed by machine learning models, and used to generate recommendations for resource reallocation. This feedback mechanism enables dynamic adjustment of resource allocation, identifying underutilized virtual machines and suggesting consolidation or de-provisioning, thereby reducing resource wastage and associated costs.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system monitors multiple parameters including CPU utilization, memory usage, storage I/O, and network traffic. By analyzing changes in these parameters over time and comparing them against thresholds and patterns, the system can identify virtual machines that are consistently underutilized across multiple metrics, enabling more accurate identification of candidates for resource optimization.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If machine learning classification is applied to all virtual machines, then accurate utilization assessment can be achieved, but the complexity of the system increases

Engineering Contradiction:

Improveutilization assessment accuracyVSAvoidsystem complexity for handling hundreds of thousands of virtual machines

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system divides the complex classification task into manageable segments by processing virtual machines in batches or groups rather than individually. It implements a hierarchical classification approach where common patterns are identified at aggregate levels, and detailed classification is applied only where needed, reducing overall system complexity while maintaining accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces intermediary components including data normalization layers, feature extraction modules, and recommendation engines that act as mediators between raw data and final classification results. These intermediaries simplify the classification process by transforming complex multi-dimensional data into standardized formats that machine learning models can process efficiently, reducing system complexity.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11055126B2Machine learning computing model for virtual machine underutilization detection

Publication Date: 2021.07.06 ROYAL BANK OF CANADA

AI summary

Systems and methods are provided for detecting sub-optimal performance of one or more virtual computing platforms. Usage data representing user activity, and performance data representing computing hardware resource utilization, is collected from a plurality of virtual machines hosted on one or more virtual computing platforms. The usage data and performance data is then analyzed along with configuration data representing the hardware components of the computing devices operating the virtual computing platform.