Network Function Virtualization Resource Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network function virtualization (NFV) systems face sub-optimal resource allocation challenges, particularly as demand patterns change, leading to inefficiencies and potential service disruptions when resources are reallocated during peak usage.
Innovation Solution
A method and apparatus for optimizing resource allocation in NFV networks by forecasting service demand, assembling candidate configurations, and assessing these configurations for optimal resource clustering, considering latency and traffic patterns, using genetic algorithms to identify the most suitable configuration that maximizes service coverage and resource utilization while minimizing delay and variance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If ad-hoc resource allocation is used to meet immediate client demands, then client service requirements are satisfied in real-time, but resource allocation becomes sub-optimal as system capacity approaches maximum
Solution Approach 1:
The patent implements predictive resource allocation by analyzing historical demand patterns and forecasting future service requirements. The system proactively allocates resources based on predicted demand before actual client requests occur, rather than reacting ad-hoc to immediate demands. This preliminary action enables optimal resource configuration in advance, preventing sub-optimal allocations that occur when resources are assigned reactively as system capacity nears maximum.
2Adaptability or versatility
If resources are reallocated on-the-fly during active sessions, then flexibility to meet changing demands is achieved, but service disruptions occur and sessions may be dropped
Solution Approach 1:
The system performs resource allocation and reconfiguration during periods of low utilization or during scheduled maintenance windows rather than during active service delivery. By predicting demand patterns and pre-configuring resources during off-peak periods, the system maintains session continuity while still achieving adaptability to changing demands. This eliminates the need for disruptive on-the-fly reallocation during active sessions.
Solution Approach 2:
The patent implements continuous monitoring of system capacity, service performance metrics, and demand patterns. This feedback mechanism enables the system to detect when resource reconfiguration is needed and to execute changes at optimal times that minimize disruption. The feedback loop allows the system to balance adaptability and reliability by making informed decisions about when and how to reconfigure resources without dropping active sessions.
3Reliability
If fixed resource configurations are used to ensure stable service delivery, then service reliability is maintained, but the system cannot adapt when demand patterns change
Solution Approach 1:
The patent transforms static resource configurations into dynamic, adaptive structures. The system continuously monitors demand patterns and automatically reconfigures resource allocations in response to changing conditions. By implementing predictive analytics and automated resource orchestration, the system maintains stability through controlled, incremental changes rather than fixed configurations, enabling both reliability and adaptability simultaneously.
Solution Approach 2:
The system employs continuous feedback mechanisms that monitor service performance, capacity utilization, and demand patterns. This feedback enables automatic adjustment of resource configurations to match changing demand while maintaining service level agreements. The feedback-driven approach ensures that adaptations are made in a controlled manner that preserves service reliability even as the system responds to evolving demand patterns.
4Productivity
If comprehensive assessment of candidate configurations is performed to identify optimal resource allocation, then resource utilization efficiency is maximized, but computational complexity and assessment time increase
Solution Approach 1:
The patent replaces complex manual or rule-based resource assessment mechanisms with machine learning algorithms and predictive analytics systems. These intelligent systems automatically evaluate multiple candidate configurations by analyzing historical data, demand patterns, and service requirements. The substitution of mechanical assessment methods with AI-driven approaches enables comprehensive evaluation of resource allocation options without proportionally increasing system complexity, as the algorithms handle the computational burden efficiently.
Solution Approach 2:
The system creates and evaluates virtual copies or simulations of potential resource configurations before implementing actual changes. By using predictive models to simulate the performance of candidate configurations based on historical data and demand patterns, the system can assess multiple options efficiently without requiring complex real-time testing. This copying approach reduces assessment complexity by working with simulated representations rather than requiring full-scale evaluation of each configuration option.
Data Source
AI summary
An optimum configuration of resources in a network function virtualisation data network is identified by assembling candidate configurations of resources (243), each configuration being an arrangement of the resources into clusters selected such that each cluster provides one or more required services, (212, 213) and assessing the candidate configurations (step 400) to identify an optimum configuration, the assessment of each configuration including measurement of latency (195) in physical links between the resources and, for each candidate configuration, determination of the total latency between the resources within each cluster of the configuration, for a predicted level and pattern of traffic associated with the required service to be operated by each cluster.