Multi-layer Stateful PCE Architecture for Network Path Computation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current software-defined networking (SDN) architectures lack a complete view of network infrastructure from physical to application layers and inefficient multi-layer computation, limiting network management and optimization capabilities.
Innovation Solution
A multi-layer, stateful Path Computation Element (PCE) architecture that determines underlying network information, topologies, and label switched path (LSP) state information, aggregating this data to establish communication among stateful computing entities for parallel computation and efficient network guidance and path computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a traditional SDN architecture is used, then network service management is simplified through abstraction, but there is no complete view of network infrastructure and no efficient multi-layer computation
Solution Approach 1:
The system segments network computation into multiple layers (physical layer, optical layer, IP/MPLS layer) with dedicated PCE instances for each layer. Each PCE maintains state information for its specific layer while communicating with other PCEs through a southbound interface, allowing distributed multi-layer computation without losing visibility into any single layer's details
Solution Approach 2:
The architecture implements a nested structure where PCEP+ extensions are embedded within the existing PCE framework. The PCEP+ protocol extends traditional PCEP by adding stateful synchronization capabilities and multi-layer coordination mechanisms, allowing the system to maintain backward compatibility while gaining enhanced multi-layer computation capabilities
2Loss of information
If network state information is aggregated across multiple layers, then comprehensive network view is achieved, but system complexity increases
Solution Approach 1:
The system divides network state management into separate PCE instances for different layers (physical PCE, optical PCE, IP/MPLS PCE). Each PCE maintains and manages state information for its specific layer independently, reducing the complexity burden on any single PCE while achieving comprehensive multi-layer visibility through coordinated state aggregation
Solution Approach 2:
The PCEP+ protocol acts as an intermediary mechanism that enables stateful synchronization between PCE instances across different layers. It provides standardized interfaces for state exchange and coordination without requiring complex custom protocols between each PCE pair, thereby managing complexity through protocol standardization
3Productivity
If parallel computation is implemented across multiple PCEs, then computation efficiency is improved, but coordination and communication overhead increases
Solution Approach 1:
The PCEP+ protocol provides universal communication capabilities that work across all PCE instances regardless of layer. It handles multiple functions including state synchronization, path computation coordination, and resource allocation through a single standardized interface, reducing communication overhead by eliminating the need for layer-specific protocols
Data Source
AI summary
In one embodiment, a stateful computing entity in a computer network determines underlying network information (physical and/or optical) for the computer network, and also determines topologies (Internet Protocol (IP) and/or Multiprotocol Label Switching (MPLS)) for the computer network and associated resource information. Further, the stateful computing entity determines label switched path (LSP) state information for the computer network. The stateful computing entity may then build network state knowledge by aggregating the underlying network information, the topologies and associated resource information, and the LSP state information, and establishes communication within a dynamic network of other stateful computing entities sharing network state knowledge for parallel computation performance. Accordingly, the stateful computing entity may perform network computation based on the network state knowledge.