Virtual Infrastructure Resource Allocation via Reference Affinity Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for managing virtual machines in data centers and cloud infrastructure face inefficiencies due to non-optimized resource utilization, leading to increased costs and service degradation from unforeseen traffic surges or reductions, and prevent dynamic resource sharing and reallocation.
Innovation Solution
A method that monitors resource utilization, calculates a reference resource affinity score (RRAS) to assess the impact of one resource unit on others, and uses this score to dynamically assign and reassign resources, enabling more precise and efficient management of virtual infrastructure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If virtual machines are migrated based on load balancing metrics, then resource utilization is improved, but system complexity and operational costs increase
Solution Approach 1:
The patent replaces complex mechanical migration operations with a predictive scoring system. Instead of physically moving virtual machines based on load balancing metrics, the system calculates a Reference Resource Affinity Score (RRAS) that predicts resource utilization patterns, allowing for simpler, more efficient resource allocation decisions without actual migration overhead
Solution Approach 2:
The patent performs preliminary analysis by calculating RRAS scores before resource allocation decisions are made. This advance prediction of resource affinity allows the system to pre-determine optimal placements, avoiding the need for reactive migrations and reducing overall system complexity
2Ease of operation
If discrete flavors are offered for virtual machines, then resource allocation is simplified, but resource under-utilization and waste increase
Solution Approach 1:
The patent introduces continuous RRAS scoring parameters that complement discrete flavor selections. By calculating affinity scores across multiple resource dimensions (CPU, memory, storage, network), the system can dynamically adjust resource allocation within flavor constraints, preventing under-utilization while maintaining operational simplicity
Solution Approach 2:
The patent implements feedback mechanisms where RRAS calculations continuously monitor resource utilization patterns. This feedback loop allows the system to identify under-utilized resources and reallocate them dynamically, reducing waste while preserving the simplicity of discrete flavor-based allocation
3Productivity
If virtual machine placement is based on predicted CPU utilization, then placement optimization is improved, but measurement precision and reliability decrease
Solution Approach 1:
The patent creates a composite measurement approach by combining multiple resource utilization metrics (CPU, memory, storage, network) into a unified RRAS score. This composite indicator provides more reliable and precise placement decisions than any single metric alone, while maintaining computational efficiency for optimization
Data Source
AI summary
A method for operating a virtual network infrastructure, wherein a corresponding physical infrastructure comprises one or more physical infrastructure resources, includes monitoring utilization levels of one or more resource units of the one or more physical infrastructure resources for virtual resources requesting the one or more resource units; calculating average absolute resource utilization values based on the utilization levels for each of the virtual resources; calculating a reference resource of score (RRAS) for each of the one or more resource units of the one or more physical infrastructure resources, wherein the RRAS indicates an impact of the utilization of a reference resource unit on utilization of other resource units on a physical infrastructure resource using the calculated average absolute resource utilization values; and assigning resources by a virtual infrastructure controller (VIC) and/or a VIC-agent on a resource, based on the RRAS for the virtual resources.