VM Migration via Prediction Module for CPU Utilization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtual machine migration technologies fail to predict and manage resource-intensive events effectively, leading to unpredictable performance and potential adverse effects on other virtual machines running on oversubscribed compute devices.
Innovation Solution
A system that includes a management device with a prediction module and an orchestration module to identify undersubscribed compute devices for virtual machine instantiation, migrate virtual machines to oversubscribed devices after booting, and predict resource-intensive events to proactively migrate virtual machines to undersubscribed devices before such events occur, thereby maintaining predictable performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If virtual machines are provisioned with more CPU's than physical CPUs (oversubscription policy), then CPU utilization of physical hardware increases, but virtual machine performance becomes unpredictable when resource-intensive events occur
Solution Approach 1:
The prediction module performs preliminary analysis of virtual machine behavior patterns to forecast resource-intensive events before they occur. This advance prediction enables the system to take preventive actions (such as migrating VMs to different hosts or allocating additional resources) before performance degradation happens, thus maintaining both high CPU utilization and predictable VM performance
Solution Approach 2:
The system implements continuous monitoring of virtual machine resource consumption and performance metrics, feeding this data back to the prediction module. This feedback loop enables dynamic adjustment of resource allocation and migration decisions, allowing the system to maintain high utilization while responding to changing conditions that could affect performance predictability
2Reliability
If virtual machines are migrated proactively before resource-intensive events, then performance predictability is maintained, but system complexity increases due to prediction and migration orchestration
Solution Approach 1:
The prediction module analyzes virtual machine behavior patterns and autonomously identifies when resource-intensive events are likely to occur, without requiring manual intervention or complex external control systems. This self-service capability simplifies the overall system architecture while maintaining performance predictability through automated prediction and triggering of migration processes
Data Source
AI summary
In some embodiments, a non-transitory processor-readable medium includes code to cause a processor to receive, at a management device, an instantiation request for a first virtual machine. The code includes code to cause the processor to identify a first compute device at a first time such that the first compute device is undersubscribed at the first time. The code includes code to cause the processor to send an instruction to instantiate the first virtual machine at the compute device, and receive a signal indicating that a boot process associated with the first virtual machine is complete and that the virtual machine is instantiated at the first compute device. The code includes code to cause the processor to send at a second time, a signal to migrate the first virtual machine from the first compute device to a second compute device in response to the boot process being complete.