GADAG-Based MRT Deployment for Layer 2 and 3 Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current solutions for implementing maximally redundant trees (MRTs) in data networks are limited to the Internet Protocol (IP) layer (Layer 3) and do not address the distribution of computation, with no existing solutions for Layer 2, and existing Layer 3 solutions are not efficient.
Innovation Solution
A method and system for computing and distributing a generalized almost directed acyclic graph (GADAG) using a path computation element (PCE) in network elements, allowing for the computation and deployment of MRTs at both Layer 2 and Layer 3, with the PCE computing a GADAG, assembling a descriptor, and distributing it to network elements for traffic forwarding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If maximally redundant trees are implemented at Layer 3 using existing solutions, then network redundancy is provided, but the computation distribution is not addressed and efficiency is reduced
Solution Approach 1:
The patent introduces a Path Computation Element (PCE) as an intermediary component that centralizes the computation of GADAG and MRT structures. The PCE receives topology information from network elements, computes the redundant path structures, and distributes the results back to network elements. This intermediary approach resolves the contradiction by providing efficient centralized computation while maintaining network-wide redundancy coverage.
Solution Approach 2:
The patent segments the MRT implementation into distinct functional components: topology information collection at network elements, centralized GADAG/MRT computation at the PCE, and distributed path selection at network elements. This segmentation allows each component to specialize in its function, improving overall computation efficiency while maintaining the reliability benefits of distributed redundancy.
2Reliability
If maximally redundant trees are implemented at Layer 2, then network robustness is enhanced, but no existing solution provides the computation distribution mechanism
Solution Approach 1:
The PCE serves as an intermediary that handles the complex computation distribution task. It collects topology information from Layer 2 network elements, computes the GADAG structure, and distributes the MRT information back to the network elements. This approach enables Layer 2 MRT implementation with automated computation distribution, reducing device complexity while maintaining enhanced network robustness.
Solution Approach 2:
The patent performs preliminary computation of the GADAG structure and MRT paths before actual network failures occur. The PCE pre-computes the redundant path structures based on current topology and distributes them to network elements in advance. When failures occur, network elements can immediately switch to pre-computed paths without performing complex real-time computations, reducing device complexity during failure scenarios.
3Device complexity
If centralized computation of GADAG is performed by a single network element, then computation is simplified, but the network element must have complete topology information
Solution Approach 1:
The PCE acts as an intermediary that collects topology information from multiple network elements through standardized protocols. Instead of requiring a single network element to have complete topology information, the PCE aggregates topology data from distributed sources, computes the GADAG structure, and distributes results back to the network. This approach simplifies computation at the PCE while eliminating the need for any single network element to maintain complete topology information.
Solution Approach 2:
The PCE is designed as a universal computation entity that can serve multiple network elements with different topology information. It implements a multi-functional role of collecting topology data from various sources, computing GADAG/MRT structures, and distributing results to different network elements. This universal approach allows centralized computation simplification while gathering complete topology information from distributed network elements.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
A method and system is disclosed to compute and distribute a generalized almost directed acyclic graph (GADAG). The method includes computing, by a network element, a GADAG for the network, where the GADAG is a graph abstraction of the network, where the plurality of network elements are the vertices of the GADAG and links connecting the plurality of network elements are the directed edges of the GADAG. The method continues with assembling a GADAG descriptor based on the computed GADAG, and distributing the GADAG descriptor to a plurality of network elements, where each of the plurality of network elements utilizes the GADAG descriptor to compute maximally redundant trees for forwarding traffic.The method also includes leveraging layer 2 attributes and features for layer 2 deployments and updating redundant trees upon topology change eventssuch that traffic outage is avoided or minimized.