VM Resource Optimization via Traffic Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for optimizing virtual machine resources in IP network systems, such as IP telephone networks, only consider the current state of resources and do not predict future traffic, leading to unstable operations and repeated live migrations that increase traffic and degrade performance.
Innovation Solution
A resource optimization method that estimates anticipated traffic based on the number of subscribers and selects physical hardware with a margin to process this traffic, allowing for proactive live migration to prevent resource shortages and optimize resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If live migration is performed based only on current resource state, then resource optimization is achieved, but traffic increases and performance deteriorates due to repeated migrations
Solution Approach 1:
The patent applies preliminary action by estimating future traffic based on the number of subscribers before performing live migration. This allows the system to proactively migrate virtual machines to physical hardware that will have sufficient capacity to handle anticipated traffic, preventing repeated migrations and maintaining system stability.
Solution Approach 2:
The patent implements feedback by continuously monitoring the number of subscribers and using this information to estimate future traffic. This feedback loop enables dynamic adjustment of migration decisions, ensuring that virtual machines are migrated to appropriate physical hardware based on actual system conditions and predicted demand.
2Ease of operation
If virtual machine is migrated to physical hardware with current resource availability, then migration is simple, but traffic immediately increases causing repeated migrations
Solution Approach 1:
The patent applies preliminary action by estimating future traffic before migration and selecting physical hardware that has sufficient capacity for the anticipated load. This prevents the scenario where migration causes immediate traffic increases, thereby avoiding repeated migrations while maintaining operational simplicity.
3Device complexity
If resource optimization is performed only when resources become scarce, then response is simple, but stable operation cannot be ensured for real-time services
Solution Approach 1:
The patent applies preliminary action by estimating future traffic based on subscriber numbers and performing live migration before resources become scarce. This proactive approach ensures stable operation for real-time services by preventing resource shortages rather than merely responding to them, while adding minimal complexity to the control mechanism.
Solution Approach 2:
The patent implements feedback by monitoring subscriber numbers and using this information to estimate future traffic and trigger migrations proactively. This feedback mechanism enables the system to maintain operation stability for real-time services without requiring complex predictive algorithms.
Data Source
AI summary
There is provided a resource optimization method enabling a stable live migration. An amount of generated traffic that is anticipated to be generated is estimated based on the number of subscribers to be processed by a SIP server on a virtual machine (step S1). Since it is not necessary to perform a live migration of the virtual machine when the estimated amount of generated traffic is less than or equal to an optimum value predetermined as throughput of physical hardware on which the virtual machine operates (“Yes” in step S2), monitoring of the number of number of subscribers is continued by returning to step S1. Meanwhile, when the estimated amount of generated traffic exceeds the optimum value (“No” in step S2), an optimum physical hardware which has a margin for processing the amount of generated traffic is selected, and the virtual machine live-migrates to the selected physical hardware (step S3).