ML Virtualization Subsystem Anticipates VM State Instability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Virtualization layers face challenges in predicting and preventing non-optimal and unstable behavior in virtual machine resource allocation, leading to sub-optimal computational throughput and operational instability when demand approaches hardware resource thresholds.
Innovation Solution
A machine-learning-based virtualization-layer subsystem that uses predictive models, such as neural networks, to anticipate potential deleterious state changes in virtual machines by analyzing resource-allocation states and adjusting resource allocation accordingly to prevent instability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the virtualization layer allocates hardware resources to virtual machines based on aggregate demand, then resource utilization is optimized, but operational stability deteriorates when demand approaches hardware thresholds
Solution Approach 1:
The machine-learning subsystem performs preliminary analysis of resource-allocation state changes before they are implemented. When a requested change is predicted to cause deleterious effects, the system proactively prevents the change from occurring, thereby maintaining system stability before instability can arise.
Solution Approach 2:
The system continuously monitors resource-allocation states and uses machine-learning models to predict the operational characteristics that would result from state changes. This feedback loop enables the virtualization layer to make informed decisions about whether to permit or prevent requested changes, balancing resource utilization with system stability.
2Adaptability or versatility
If the virtualization layer permits state changes to virtual machines, then adaptability improves, but system stability deteriorates when changes lead to non-optimal operational characteristics
Solution Approach 1:
The machine-learning-based subsystem acts as an intermediary between resource-allocation requests and the virtualization layer's state changes. This intermediary analyzes predicted operational characteristics and selectively permits or prevents changes, enabling the system to adapt to new requirements while filtering out changes that would cause instability.
3Reliability
If system monitoring is increased to prevent non-optimal behavior, then operational stability improves, but system complexity increases
Solution Approach 1:
The machine-learning subsystem enables the virtualization layer to self-monitor and self-regulate resource allocation decisions. By automatically analyzing predicted operational characteristics and making preventive decisions, the system achieves enhanced stability without requiring complex external monitoring infrastructure or manual intervention.
Data Source
AI summary
The current document is directed to a machine-learning-based subsystem, included within a virtualization layer, that learns, over time, how to accurately predict operational characteristics for the virtual machines executing within the virtual execution environment provided by the virtualization layer that result from changes to the states of the virtual machines. When the virtualization layer receives requests that, if satisfied, would result in a change of the state of one or more virtual machines, the virtualization layer uses operational characteristics predicted by the machine-learning-based subsystem from virtual-machine resource-allocation states that would obtain by satisfying the requests. When the predicted operational characteristics are indicative of potential non-optimality, instability, or unpredictability of virtualized-computer-system operation, the virtualization layer anticipates a deleterious state change and undertakes preventative measures.