Unified WAN Label Routing for Scalable Traffic Engineering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The operation of separate software-defined WAN (SWAN) and standards-defined WAN (CORE) networks for inter-datacenter and Internet traffic is complex, costly, and inefficient, leading to capacity planning difficulties and wasteful use of network equipment, with the RSVP-TE protocol reaching scale limits and BGP requiring large TCAMs.
Innovation Solution
A unified wide area network (WAN) architecture with aggregation routers holding full IP routing tables and backbone routers as forwarding-only devices, using MPLS for encapsulation and traffic engineering to route both inter-datacenter and Internet traffic, employing traffic engineered tunnels and node segment identifiers for optimized routing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If separate SWAN and CORE networks are used for inter-datacenter and Internet traffic, then traffic routing is possible, but network complexity and cost increase
Solution Approach 1:
The patent merges SWAN and CORE networks into a unified WAN fabric, allowing both inter-datacenter and Internet traffic to traverse a single network infrastructure. This consolidation eliminates the need for separate network planes while maintaining distinct traffic engineering capabilities through label-based routing.
Solution Approach 2:
The unified WAN architecture provides multi-functionality by enabling a single network to handle both dedicated inter-datacenter traffic and general Internet traffic. The system uses MPLS labels to differentiate traffic types and apply appropriate routing policies, making the network universally applicable to multiple traffic scenarios.
2Productivity
If RSVP-TE protocol is used for traffic engineering, then optimized routing is achieved, but scale limits are reached
Solution Approach 1:
The patent replaces the RSVP-TE signaling mechanism with MPLS label-based routing. Instead of using complex reservation protocols, the system uses simplified label switching to achieve traffic engineering goals, thereby scaling to larger network sizes without the protocol overhead and state management complexity of RSVP-TE.
3Ease of operation
If BGP is used for routing, then Internet traffic is routed, but large TCAM requirements arise
Solution Approach 1:
The patent extracts the routing decision function from backbone routers and places it in edge routers. Backbone routers only perform simple label switching based on pre-computed paths, eliminating the need for large TCAMs. Edge routers maintain the full routing tables and make routing decisions, while backbone routers handle high-speed forwarding with minimal memory requirements.
4Measurement precision
If aggregation routers hold full IP routing tables, then routing decisions are accurate, but backbone router complexity increases
Solution Approach 1:
The patent segments the routing function across different router types. Aggregation routers hold full IP routing tables and make routing decisions, while backbone routers are segmented to perform only label-based forwarding. This segmentation allows accurate routing decisions at the aggregation layer while keeping backbone routers simple and efficient.
Data Source
AI summary
A method and a network for routing data packet in a unified wide area network (WAN) is provided. The method includes encapsulating a data packet by an ingress aggregation router and forwarding the encapsulated data packet to an ingress backbone router. The encapsulated data packet includes a first label. The ingress backbone router selects an optimized traffic engineered tunnel and replaces the first label with the optimized traffic engineered tunnel and forwards the encapsulated data packet along the optimized traffic engineered tunnel.