Virtual Workload Priority Management via Preemption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In computing environments, managing virtual workload priorities is challenging when physical resources are limited, as determining which virtual machines to remove or modify to deploy new ones is difficult due to varying resource requirements over a virtual machine's lifecycle.
Innovation Solution
A method that determines when preemption is required, identifies hosts with preemptible virtual machines, prioritizes them based on resource usage, and selects hosts to deploy new virtual machines by stopping or modifying preemptible virtual machines to allocate resources efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If virtualization is employed to use resources more efficiently, then resource utilization improves, but difficulties arise in determining which virtual machines to remove when physical resources are limited
Solution Approach 1:
The system changes the parameter of virtual machine priority levels, categorizing them into preemptible and non-preemptible classes. This parameter change enables automated decision-making by assigning different removal priorities based on workload importance, thereby resolving the complexity of determining which virtual machines to remove while maintaining efficient resource utilization.
2Productivity
If preemptible virtual machines are stopped to deploy new virtual machines, then resource allocation efficiency improves, but disruptions to existing workloads occur
Solution Approach 1:
The system applies local quality by differentiating between preemptible and non-preemptible virtual machines based on their specific workload characteristics. Critical workloads are marked as non-preemptible to maintain continuity, while less critical workloads are marked as preemptible to enable flexible resource allocation. This localized quality assignment resolves the contradiction by allowing disruptions only where acceptable.
Solution Approach 2:
The system implements dynamic priority assignment where virtual machines can be classified as preemptible or non-preemptible based on changing workload requirements. This dynamic classification allows the system to adapt resource allocation strategies in real-time, enabling efficient resource reallocation when disruptions are acceptable while protecting critical workloads when continuity is essential.
3Productivity
If resource shares are increased for a virtual machine, then performance of that virtual machine improves, but resource availability for other virtual machines decreases
Solution Approach 1:
The system applies partial action by selectively increasing resource shares only for non-preemptible virtual machines that require additional resources, while allowing preemptible virtual machines to operate with reduced or unchanged resource allocation. This partial resource reallocation improves the performance of critical workloads without completely depriving other virtual machines of necessary resources.
Data Source
AI summary
Described herein are systems, methods, and software to manage virtual workload priorities in a computing environment. In at least one example, a management service determines when preemption is required to deploy a virtual machine in a computing environment. When preemption is required, the management service identifies one or more hosts in the computing environment capable of supporting the virtual machine when at least one preemptible virtual machine of one or more preemptible virtual machines is stopped on a corresponding host of the one or more hosts. The management service then selects a host from the one or more hosts for the virtual machine, stops the at least one virtual machine, and deploys the virtual machine on the host.