Decoupling Reservation and Data Forwarding Paths in Network Routers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current reservation protocols in computer networks are inefficient as they couple reservation and data forwarding paths, leading to wastage of resources on non-shortest paths and increased processing load, especially when dynamic redistribution of data flows is required.
Innovation Solution
A technique that decouples reservation and data forwarding by allowing reservations on paths with adequate reservation availability and forwarding data on paths with suitable load availability, using a directed acyclic graph (DAG) to determine the best paths based on metrics like cost, load, and reservation availability, without requiring per-flow states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If reservation and data forwarding are coupled on the same path, then reservation availability is ensured, but network resource utilization deteriorates and processing load increases
Solution Approach 1:
The patent segments the reservation function and data forwarding function into two independent paths. Reservation messages traverse a reserved path established through signaling protocols, while data messages traverse a forwarding path determined by load conditions. This segmentation allows each function to optimize independently, resolving the contradiction between ensuring reservation availability and maximizing network resource utilization.
Solution Approach 2:
The patent extracts the data forwarding decision from the reservation path constraints. By allowing data to be forwarded on paths different from the reserved path, the system takes out the restriction that previously coupled reservation and forwarding. This enables data to follow the shortest or least-loaded path while reservations maintain their established paths, improving overall network utilization without compromising reservation reliability.
2Reliability
If reservations are made on non-shortest paths to ensure availability, then reservation success rate improves, but network efficiency deteriorates
Solution Approach 1:
The patent separates the reservation path selection criterion from the data forwarding path selection criterion. Reservation paths are selected based on availability and QoS guarantees, while data forwarding paths are selected independently based on shortest path or load-based metrics. This segmentation allows reservations to succeed on appropriate paths while data transmission occurs on optimal paths, resolving the contradiction between reservation success rate and transmission efficiency.
3Adaptability or versatility
If per-flow states are maintained for dynamic redistribution, then routing flexibility improves, but device complexity and processing load increase
Solution Approach 1:
The patent extracts the flow state information from individual routers and centralizes it in a Path Computation Element (PCE). The PCE receives path computation requests, calculates optimal paths using global network state information, and returns the computed paths. This extraction eliminates the need for each router to maintain complex per-flow states, reducing device complexity and processing load while preserving routing flexibility through centralized intelligent control.
Solution Approach 2:
The patent introduces a Path Computation Element (PCE) as an intermediary between the network nodes and the path computation function. The PCE acts as a mediator that collects global network state information, performs path computations for multiple flows, and distributes the computed paths to relevant nodes. This intermediary approach provides routing flexibility without requiring complex per-flow state maintenance at individual routers, as the PCE handles the computational complexity centrally.
Data Source
AI summary
In one embodiment, a node (e.g., a router) performs reservations for data flows, each on a corresponding selected (reserved) path having adequate reservation availability. Also, the node forwards data from data flows, each over a corresponding selected (forwarded) path having adequate load availability, wherein forwarded paths are decoupled from reserved paths for the data flows.