Virtual Network Function Placement Using Historical Performance Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed cloud environments, existing methods for placing virtual network functions (VNFs) are inefficient due to reliance on current infrastructure information, which fails to account for actual performance, leading to inaccurate placement decisions and increased complexity in provisioning new services.
Innovation Solution
A method that uses historical data to calculate similarity values between candidate nodes and previous VNF placements, determining the most suitable node for VNF instantiation based on performance indicators, thereby reducing dependency on resource descriptors and improving placement accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If placement decisions are based solely on current infrastructure information, then the placement process is simple and fast, but the placement accuracy is poor
Solution Approach 1:
The system performs preliminary actions by collecting and storing historical placement data and performance indicators before making new placement decisions. This historical data is prepared in advance and can be quickly referenced when a new VNF needs to be placed, improving accuracy without adding complex real-time analysis
Solution Approach 2:
The system implements feedback by using historical performance indicators from previous VNF placements to inform and improve future placement decisions. The performance data feeds back into the placement algorithm, creating a continuous improvement loop that enhances accuracy over time
2Measurement precision
If historical data is used for placement decisions, then placement accuracy is improved, but the computational complexity increases
Solution Approach 1:
The system extracts only the most relevant features and performance indicators from historical data that are necessary for making accurate placement decisions. By selecting and using only the critical data elements rather than processing all historical information, the system maintains high accuracy while reducing computational overhead
Solution Approach 2:
The system applies partial action by using a subset of historical data and performance metrics that are most predictive of successful placements. Rather than exhaustively analyzing all possible historical parameters, the system focuses on the most influential factors, achieving good results with reduced computational effort
3Measurement precision
If more candidate nodes are evaluated, then the chance of finding an optimal placement increases, but the placement time increases
Solution Approach 1:
The system performs preliminary evaluation of candidate nodes using historical performance data before the actual placement decision. Nodes are pre-ranked or pre-filtered based on their historical performance indicators, so that when a new VNF needs placement, the system can quickly identify the most promising candidates without evaluating all nodes in detail
Solution Approach 2:
The system evaluates only a partial set of candidate nodes in detail, focusing computational resources on the most promising candidates identified through historical data analysis. By concentrating evaluation efforts on a subset of high-potential nodes rather than exhaustively evaluating all candidates, the system achieves good placement optimality within acceptable timeframes
Data Source
AI summary
A method for placing a virtual network function (VNF) in a datacenter system based on historical data. The method includes receiving, by a placement service from a client device, a VNF placement request that includes an identifier of the VNF, wherein the VNF placement request requests placement of an instantiation of the VNF with a node in a set of candidate nodes of the datacenter system; calculating, by the placement service, a set of similarity values between the set of candidate nodes of the datacenter system and the historical data, wherein the historical data represents previous placements of the VNF in the datacenter system, including performance indicators associated with these placements; and determining, by the placement service, a placement node from the set of candidate nodes for placing the instantiation of the VNF based on the set of similarity values.