Virtual Machine Migration Response Time Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional virtual machine management techniques struggle to accurately predict response time improvements when overcommitting resources, as they fail to account for the risk of resource competition between virtual machines sharing resources on the same physical machine, leading to potential performance issues during increased load times.
Innovation Solution
A computer-readable storage medium that calculates the non-shared resource availability on a target physical machine and uses different algorithms to estimate response time based on whether the virtual machine uses non-shared or shared resources, accounting for potential competition and resource usage scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If overcommit is performed to improve resource usage efficiency, then resource efficiency is improved, but response time reliability deteriorates because virtual machines cannot guarantee usage of shared resources
Solution Approach 1:
The patent segments the resource assignment into two distinct types: non-shared resources (guaranteed to the virtual machine) and shared resources (potentially available but not guaranteed). This segmentation allows the system to maintain overcommit for efficiency while clearly distinguishing between guaranteed and potential resource usage, thereby resolving the contradiction between resource efficiency and response time reliability.
Solution Approach 2:
The patent changes the parameter of resource allocation by introducing a new dimension - the guarantee level of resource usage. Instead of a single uniform resource allocation, the system now allocates resources with different guarantee characteristics (guaranteed vs. non-guaranteed), enabling the system to optimize both efficiency and reliability by selecting appropriate allocation strategies based on workload requirements.
2Device complexity
If conventional response time estimation algorithms are used without considering shared resources, then calculation simplicity is maintained, but estimation accuracy deteriorates because they fail to account for resource competition risks
Solution Approach 1:
The patent segments the response time estimation process into two distinct estimation paths: one for non-shared resources and another for shared resources. Each path uses appropriate algorithms suited to its specific characteristics, with the shared resource path accounting for competition risks. This segmentation maintains overall algorithm simplicity while significantly improving estimation accuracy by handling different resource scenarios appropriately.
Data Source
AI summary
An information processing apparatus calculates, at the time of moving a virtual machine which operates on a first physical machine, an amount of a first resource which the virtual machine can use on a second physical machine without sharing with another virtual machine. The information processing apparatus estimates response time of the virtual machine at the time of moving the virtual machine to the second physical machine on the basis of response time of the virtual machine at the time of the virtual machine operating on the first physical machine and the amount of the first resource. At this time the information processing apparatus uses different algorithms at the time of a resource used by the virtual machine being within the first resource and at the time of the virtual machine using a second resource which the virtual machine shares with another virtual machine.