Programmable Graph Model for Software-Defined Network Introspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional networks lack programmability and introspection, making it impractical to configure them dynamically in response to real-time changes, as intelligence is external and not intrinsic to the network hardware.
Innovation Solution
A programmable graph model is used to represent software-defined networks, enabling distributed operation management across network elements, providing introspection and reflection capabilities through a common programming model and APIs, allowing for dynamic configuration and function application.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If networks are configured at individual devices using traditional programming frameworks, then network operations can be implemented, but the intelligence is external to the network and not intrinsic, making it impractical for complex networks to respond to real-time changes
Solution Approach 1:
The patent embeds programmable intelligence within network switches and routers themselves, creating nested layers of intelligence where individual devices contain embedded controllers that can execute operations locally. This nested structure allows the network to respond to changes autonomously at the device level while maintaining overall system coordination.
Solution Approach 2:
The patent introduces a graph-based intermediate representation that serves as a mediator between high-level network operations and low-level device configurations. This graph model acts as an intermediary layer that translates operational intent into device-specific configurations, simplifying the complexity of individual device configuration while maintaining adaptability.
2Adaptability or versatility
If traditional network hardware is used, then network operations can be performed, but the hardware lacks introspection and reflection capabilities, preventing dynamic manipulation of network element properties
Solution Approach 1:
The patent enables network switches and routers to perform self-introspection by embedding controllers that can examine their own state, properties, and operational parameters. This self-service capability allows devices to provide introspection information about themselves without requiring external probing, enabling dynamic manipulation of network element properties through programmatic access.
Solution Approach 2:
The patent replaces traditional mechanical/configuration-based network management with software-based introspection and reflection capabilities. Instead of physical access or manual configuration, the system uses embedded controllers and graph-based programming to dynamically examine and manipulate network element properties through software interfaces.
3Adaptability or versatility
If more programmability is added to each switch and router, then network operations can be implemented more flexibly, but the complexity of managing individual device configurations increases
Solution Approach 1:
The patent creates a universal graph-based programming model that can operate across diverse network devices regardless of their specific hardware or software implementations. This universal interface allows the same high-level operations to be applied to different device types, reducing configuration management complexity while maintaining high programmability at the individual device level.
Solution Approach 2:
The patent transitions from managing network configuration at the individual device dimension to managing it at the network-wide graph dimension. By representing the network as a graph where nodes and edges can be manipulated collectively, the system enables operations that span multiple devices simultaneously, reducing the complexity of managing individual configurations while maintaining flexibility.
Data Source
AI summary
System, method, and computer program product to represent a network using a programmable graph model, by generating a directed graph to represent a topology of the network, wherein each of a plurality of network elements in the network are represented, in the directed graph, by one of the plurality of nodes, identifying, through the directed graph, a subset of network elements, of the plurality of network elements, upon which to apply a requested operation, and applying the requested operation to the subset of network elements in a distributed manner through the directed graph.