Synthetic Network Topologies for Graph Workload Placement Testing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in efficiently testing graph-based workload placement algorithms at scale and in controlled scenarios due to the manual and error-prone process of creating large infrastructure simulations, with public datasets being difficult to tailor for specific workloads and infrastructures.
Innovation Solution
A graph-growing approach that starts from graph-based workload seeds, determining the number of layers and nodes to add, connectivity rates, and resource availability, allowing for the creation of synthetic network infrastructures tailored for workload placement evaluations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If manual methods are used to define large infrastructure for simulation, then flexibility in creating custom scenarios is possible, but the process becomes daunting and error-prone
Solution Approach 1:
The patent uses public infrastructure datasets as templates or copies to generate synthetic infrastructure simulations. Instead of manually defining each infrastructure component, the system replicates and adapts existing infrastructure patterns, automatically scaling them to create large simulation environments while maintaining structural accuracy and reducing manual effort
Solution Approach 2:
The system pre-processes public infrastructure datasets to extract topology patterns, node types, and connection structures before generating synthetic simulations. This preliminary analysis creates reusable infrastructure templates that can be quickly instantiated and customized, eliminating the need to manually define infrastructure from scratch for each simulation scenario
2Quantity of substance
If public datasets are used for simulation, then available infrastructure data can be leveraged, but tailoring them for specific workloads and infrastructures becomes very difficult
Solution Approach 1:
The patent applies local quality by selectively adapting specific portions of public infrastructure datasets rather than uniformly transforming entire datasets. The system identifies and modifies only the topology patterns, node attributes, and connection structures relevant to specific workload requirements, leaving other aspects unchanged. This allows precise tailoring of infrastructure simulations to match specific workload characteristics while preserving the overall structure and scale of public datasets
3Manufacturing precision
If manual infrastructure definition is used, then control over simulation parameters is possible, but the process is time-consuming and prone to errors
Solution Approach 1:
The system implements feedback mechanisms that automatically validate generated infrastructure simulations against desired parameter specifications. During the automated generation process from public datasets, the system continuously checks whether topology patterns, node distributions, and connection structures meet the required simulation parameters, making real-time adjustments to maintain precision while preserving high productivity through automation
Data Source
AI summary
Techniques for simulating and testing infrastructure are disclosed. A service accesses a first workload comprising a set of tasks. The service builds a synthetic network infrastructure based on the first workload. The service finds at least one sub-graph within the synthetic network topology that has a same topology as the first seed graph. The service then allocates network resources associated with the at least one sub-graph to test the first workload.